66
Switching in LANs Lecture 5 CS 653, Fall 2008
| Date post: | 26-Dec-2015 |
| Category: |
Documents |
| Upload: | lydia-daniels |
| View: | 253 times |
| Download: | 1 times |
Switching in LANs
Lecture 5CS 653 Fall 2008
Link layer addressing
Adaptors Communicating
link layer implemented in ldquoadaptorrdquo (aka NIC) Ethernet card 80211
card sending side
encapsulates datagram in a frame
adds error checking bits rdt flow control etc
receiving side looks for errors rdt
flow control etc extracts datagram
passes to rcving node adapter is semi-
autonomous link amp physical layers
sendingnode
frame
rcvingnode
datagram
frame
adapter adapter
link layer protocol
MAC Addresses
MAC = Media Access Control All stations receive all packets Only keep packets for our address or
explicit broadcast packets
MAC Addresses and ARP
32-bit IP address network-layer address used to get datagram to destination IP
subnet
MAC (or LAN or physical or Ethernet) address used to get frame from one interface to
another physically-connected interface (same network)
48 bit IEEE MAC address (for most LANs) burned in the adapter ROM
LAN Addresses and ARP
Each adapter on LAN has unique LAN address
Broadcast address =FF-FF-FF-FF-FF-FF
= adapter
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN(wired orwireless)
ARP Address Resolution Protocol
Each IP node (Host Router) on LAN has ARP table
ARP Table IPMAC address mappings for some LAN nodes
lt IP address MAC address TTLgt TTL (Time To Live)
time after which address mapping will be forgotten (typically 20 min)
Question how to determineMAC address of Bknowing Brsquos IP address
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN
237196723
237196778
237196714
237196788
ARP protocol Same LAN (network)
A wants to send datagram to B and Brsquos MAC address not in Arsquos ARP table
A broadcasts ARP query packet containing Bs IP address Dest MAC address =
FF-FF-FF-FF-FF-FF all machines on LAN
receive ARP query B receives ARP packet
replies to A with its (Bs) MAC address frame sent to Arsquos MAC
address (unicast)
A caches (saves) IP-to-MAC address pair in its ARP table until information becomes old (times out) soft state information
that times out (goes away) unless refreshed
ARP is ldquoplug-and-playrdquo nodes create their ARP
tables without intervention from net administrator
Plug-and-play an incredibly nice property
ARP trace
Frame 203 (42 bytes on wire 42 bytes captured)Address Resolution Protocol (request) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode request (0x0001) Sender MAC address DellComp_5e40b9 (00065b5e40b9) Sender IP address 12811924581 (12811924581) Target MAC address 000000_000000 (000000000000) Target IP address 128119245254 (128119245254)
Frame 204 (60 bytes on wire 60 bytes captured)Address Resolution Protocol (reply) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode reply (0x0002) Sender MAC address DigitalE_00e80b (aa000400e80b) Sender IP address 128119245254 (128119245254) Target MAC address DellComp_5e40b9 (00065b5e40b9) Target IP address 12811924581 (12811924581)
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Link layer addressing
Adaptors Communicating
link layer implemented in ldquoadaptorrdquo (aka NIC) Ethernet card 80211
card sending side
encapsulates datagram in a frame
adds error checking bits rdt flow control etc
receiving side looks for errors rdt
flow control etc extracts datagram
passes to rcving node adapter is semi-
autonomous link amp physical layers
sendingnode
frame
rcvingnode
datagram
frame
adapter adapter
link layer protocol
MAC Addresses
MAC = Media Access Control All stations receive all packets Only keep packets for our address or
explicit broadcast packets
MAC Addresses and ARP
32-bit IP address network-layer address used to get datagram to destination IP
subnet
MAC (or LAN or physical or Ethernet) address used to get frame from one interface to
another physically-connected interface (same network)
48 bit IEEE MAC address (for most LANs) burned in the adapter ROM
LAN Addresses and ARP
Each adapter on LAN has unique LAN address
Broadcast address =FF-FF-FF-FF-FF-FF
= adapter
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN(wired orwireless)
ARP Address Resolution Protocol
Each IP node (Host Router) on LAN has ARP table
ARP Table IPMAC address mappings for some LAN nodes
lt IP address MAC address TTLgt TTL (Time To Live)
time after which address mapping will be forgotten (typically 20 min)
Question how to determineMAC address of Bknowing Brsquos IP address
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN
237196723
237196778
237196714
237196788
ARP protocol Same LAN (network)
A wants to send datagram to B and Brsquos MAC address not in Arsquos ARP table
A broadcasts ARP query packet containing Bs IP address Dest MAC address =
FF-FF-FF-FF-FF-FF all machines on LAN
receive ARP query B receives ARP packet
replies to A with its (Bs) MAC address frame sent to Arsquos MAC
address (unicast)
A caches (saves) IP-to-MAC address pair in its ARP table until information becomes old (times out) soft state information
that times out (goes away) unless refreshed
ARP is ldquoplug-and-playrdquo nodes create their ARP
tables without intervention from net administrator
Plug-and-play an incredibly nice property
ARP trace
Frame 203 (42 bytes on wire 42 bytes captured)Address Resolution Protocol (request) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode request (0x0001) Sender MAC address DellComp_5e40b9 (00065b5e40b9) Sender IP address 12811924581 (12811924581) Target MAC address 000000_000000 (000000000000) Target IP address 128119245254 (128119245254)
Frame 204 (60 bytes on wire 60 bytes captured)Address Resolution Protocol (reply) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode reply (0x0002) Sender MAC address DigitalE_00e80b (aa000400e80b) Sender IP address 128119245254 (128119245254) Target MAC address DellComp_5e40b9 (00065b5e40b9) Target IP address 12811924581 (12811924581)
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Adaptors Communicating
link layer implemented in ldquoadaptorrdquo (aka NIC) Ethernet card 80211
card sending side
encapsulates datagram in a frame
adds error checking bits rdt flow control etc
receiving side looks for errors rdt
flow control etc extracts datagram
passes to rcving node adapter is semi-
autonomous link amp physical layers
sendingnode
frame
rcvingnode
datagram
frame
adapter adapter
link layer protocol
MAC Addresses
MAC = Media Access Control All stations receive all packets Only keep packets for our address or
explicit broadcast packets
MAC Addresses and ARP
32-bit IP address network-layer address used to get datagram to destination IP
subnet
MAC (or LAN or physical or Ethernet) address used to get frame from one interface to
another physically-connected interface (same network)
48 bit IEEE MAC address (for most LANs) burned in the adapter ROM
LAN Addresses and ARP
Each adapter on LAN has unique LAN address
Broadcast address =FF-FF-FF-FF-FF-FF
= adapter
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN(wired orwireless)
ARP Address Resolution Protocol
Each IP node (Host Router) on LAN has ARP table
ARP Table IPMAC address mappings for some LAN nodes
lt IP address MAC address TTLgt TTL (Time To Live)
time after which address mapping will be forgotten (typically 20 min)
Question how to determineMAC address of Bknowing Brsquos IP address
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN
237196723
237196778
237196714
237196788
ARP protocol Same LAN (network)
A wants to send datagram to B and Brsquos MAC address not in Arsquos ARP table
A broadcasts ARP query packet containing Bs IP address Dest MAC address =
FF-FF-FF-FF-FF-FF all machines on LAN
receive ARP query B receives ARP packet
replies to A with its (Bs) MAC address frame sent to Arsquos MAC
address (unicast)
A caches (saves) IP-to-MAC address pair in its ARP table until information becomes old (times out) soft state information
that times out (goes away) unless refreshed
ARP is ldquoplug-and-playrdquo nodes create their ARP
tables without intervention from net administrator
Plug-and-play an incredibly nice property
ARP trace
Frame 203 (42 bytes on wire 42 bytes captured)Address Resolution Protocol (request) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode request (0x0001) Sender MAC address DellComp_5e40b9 (00065b5e40b9) Sender IP address 12811924581 (12811924581) Target MAC address 000000_000000 (000000000000) Target IP address 128119245254 (128119245254)
Frame 204 (60 bytes on wire 60 bytes captured)Address Resolution Protocol (reply) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode reply (0x0002) Sender MAC address DigitalE_00e80b (aa000400e80b) Sender IP address 128119245254 (128119245254) Target MAC address DellComp_5e40b9 (00065b5e40b9) Target IP address 12811924581 (12811924581)
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
MAC Addresses
MAC = Media Access Control All stations receive all packets Only keep packets for our address or
explicit broadcast packets
MAC Addresses and ARP
32-bit IP address network-layer address used to get datagram to destination IP
subnet
MAC (or LAN or physical or Ethernet) address used to get frame from one interface to
another physically-connected interface (same network)
48 bit IEEE MAC address (for most LANs) burned in the adapter ROM
LAN Addresses and ARP
Each adapter on LAN has unique LAN address
Broadcast address =FF-FF-FF-FF-FF-FF
= adapter
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN(wired orwireless)
ARP Address Resolution Protocol
Each IP node (Host Router) on LAN has ARP table
ARP Table IPMAC address mappings for some LAN nodes
lt IP address MAC address TTLgt TTL (Time To Live)
time after which address mapping will be forgotten (typically 20 min)
Question how to determineMAC address of Bknowing Brsquos IP address
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN
237196723
237196778
237196714
237196788
ARP protocol Same LAN (network)
A wants to send datagram to B and Brsquos MAC address not in Arsquos ARP table
A broadcasts ARP query packet containing Bs IP address Dest MAC address =
FF-FF-FF-FF-FF-FF all machines on LAN
receive ARP query B receives ARP packet
replies to A with its (Bs) MAC address frame sent to Arsquos MAC
address (unicast)
A caches (saves) IP-to-MAC address pair in its ARP table until information becomes old (times out) soft state information
that times out (goes away) unless refreshed
ARP is ldquoplug-and-playrdquo nodes create their ARP
tables without intervention from net administrator
Plug-and-play an incredibly nice property
ARP trace
Frame 203 (42 bytes on wire 42 bytes captured)Address Resolution Protocol (request) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode request (0x0001) Sender MAC address DellComp_5e40b9 (00065b5e40b9) Sender IP address 12811924581 (12811924581) Target MAC address 000000_000000 (000000000000) Target IP address 128119245254 (128119245254)
Frame 204 (60 bytes on wire 60 bytes captured)Address Resolution Protocol (reply) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode reply (0x0002) Sender MAC address DigitalE_00e80b (aa000400e80b) Sender IP address 128119245254 (128119245254) Target MAC address DellComp_5e40b9 (00065b5e40b9) Target IP address 12811924581 (12811924581)
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
MAC Addresses and ARP
32-bit IP address network-layer address used to get datagram to destination IP
subnet
MAC (or LAN or physical or Ethernet) address used to get frame from one interface to
another physically-connected interface (same network)
48 bit IEEE MAC address (for most LANs) burned in the adapter ROM
LAN Addresses and ARP
Each adapter on LAN has unique LAN address
Broadcast address =FF-FF-FF-FF-FF-FF
= adapter
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN(wired orwireless)
ARP Address Resolution Protocol
Each IP node (Host Router) on LAN has ARP table
ARP Table IPMAC address mappings for some LAN nodes
lt IP address MAC address TTLgt TTL (Time To Live)
time after which address mapping will be forgotten (typically 20 min)
Question how to determineMAC address of Bknowing Brsquos IP address
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN
237196723
237196778
237196714
237196788
ARP protocol Same LAN (network)
A wants to send datagram to B and Brsquos MAC address not in Arsquos ARP table
A broadcasts ARP query packet containing Bs IP address Dest MAC address =
FF-FF-FF-FF-FF-FF all machines on LAN
receive ARP query B receives ARP packet
replies to A with its (Bs) MAC address frame sent to Arsquos MAC
address (unicast)
A caches (saves) IP-to-MAC address pair in its ARP table until information becomes old (times out) soft state information
that times out (goes away) unless refreshed
ARP is ldquoplug-and-playrdquo nodes create their ARP
tables without intervention from net administrator
Plug-and-play an incredibly nice property
ARP trace
Frame 203 (42 bytes on wire 42 bytes captured)Address Resolution Protocol (request) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode request (0x0001) Sender MAC address DellComp_5e40b9 (00065b5e40b9) Sender IP address 12811924581 (12811924581) Target MAC address 000000_000000 (000000000000) Target IP address 128119245254 (128119245254)
Frame 204 (60 bytes on wire 60 bytes captured)Address Resolution Protocol (reply) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode reply (0x0002) Sender MAC address DigitalE_00e80b (aa000400e80b) Sender IP address 128119245254 (128119245254) Target MAC address DellComp_5e40b9 (00065b5e40b9) Target IP address 12811924581 (12811924581)
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
LAN Addresses and ARP
Each adapter on LAN has unique LAN address
Broadcast address =FF-FF-FF-FF-FF-FF
= adapter
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN(wired orwireless)
ARP Address Resolution Protocol
Each IP node (Host Router) on LAN has ARP table
ARP Table IPMAC address mappings for some LAN nodes
lt IP address MAC address TTLgt TTL (Time To Live)
time after which address mapping will be forgotten (typically 20 min)
Question how to determineMAC address of Bknowing Brsquos IP address
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN
237196723
237196778
237196714
237196788
ARP protocol Same LAN (network)
A wants to send datagram to B and Brsquos MAC address not in Arsquos ARP table
A broadcasts ARP query packet containing Bs IP address Dest MAC address =
FF-FF-FF-FF-FF-FF all machines on LAN
receive ARP query B receives ARP packet
replies to A with its (Bs) MAC address frame sent to Arsquos MAC
address (unicast)
A caches (saves) IP-to-MAC address pair in its ARP table until information becomes old (times out) soft state information
that times out (goes away) unless refreshed
ARP is ldquoplug-and-playrdquo nodes create their ARP
tables without intervention from net administrator
Plug-and-play an incredibly nice property
ARP trace
Frame 203 (42 bytes on wire 42 bytes captured)Address Resolution Protocol (request) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode request (0x0001) Sender MAC address DellComp_5e40b9 (00065b5e40b9) Sender IP address 12811924581 (12811924581) Target MAC address 000000_000000 (000000000000) Target IP address 128119245254 (128119245254)
Frame 204 (60 bytes on wire 60 bytes captured)Address Resolution Protocol (reply) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode reply (0x0002) Sender MAC address DigitalE_00e80b (aa000400e80b) Sender IP address 128119245254 (128119245254) Target MAC address DellComp_5e40b9 (00065b5e40b9) Target IP address 12811924581 (12811924581)
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
ARP Address Resolution Protocol
Each IP node (Host Router) on LAN has ARP table
ARP Table IPMAC address mappings for some LAN nodes
lt IP address MAC address TTLgt TTL (Time To Live)
time after which address mapping will be forgotten (typically 20 min)
Question how to determineMAC address of Bknowing Brsquos IP address
1A-2F-BB-76-09-AD
58-23-D7-FA-20-B0
0C-C4-11-6F-E3-98
71-65-F7-2B-08-53
LAN
237196723
237196778
237196714
237196788
ARP protocol Same LAN (network)
A wants to send datagram to B and Brsquos MAC address not in Arsquos ARP table
A broadcasts ARP query packet containing Bs IP address Dest MAC address =
FF-FF-FF-FF-FF-FF all machines on LAN
receive ARP query B receives ARP packet
replies to A with its (Bs) MAC address frame sent to Arsquos MAC
address (unicast)
A caches (saves) IP-to-MAC address pair in its ARP table until information becomes old (times out) soft state information
that times out (goes away) unless refreshed
ARP is ldquoplug-and-playrdquo nodes create their ARP
tables without intervention from net administrator
Plug-and-play an incredibly nice property
ARP trace
Frame 203 (42 bytes on wire 42 bytes captured)Address Resolution Protocol (request) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode request (0x0001) Sender MAC address DellComp_5e40b9 (00065b5e40b9) Sender IP address 12811924581 (12811924581) Target MAC address 000000_000000 (000000000000) Target IP address 128119245254 (128119245254)
Frame 204 (60 bytes on wire 60 bytes captured)Address Resolution Protocol (reply) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode reply (0x0002) Sender MAC address DigitalE_00e80b (aa000400e80b) Sender IP address 128119245254 (128119245254) Target MAC address DellComp_5e40b9 (00065b5e40b9) Target IP address 12811924581 (12811924581)
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
ARP protocol Same LAN (network)
A wants to send datagram to B and Brsquos MAC address not in Arsquos ARP table
A broadcasts ARP query packet containing Bs IP address Dest MAC address =
FF-FF-FF-FF-FF-FF all machines on LAN
receive ARP query B receives ARP packet
replies to A with its (Bs) MAC address frame sent to Arsquos MAC
address (unicast)
A caches (saves) IP-to-MAC address pair in its ARP table until information becomes old (times out) soft state information
that times out (goes away) unless refreshed
ARP is ldquoplug-and-playrdquo nodes create their ARP
tables without intervention from net administrator
Plug-and-play an incredibly nice property
ARP trace
Frame 203 (42 bytes on wire 42 bytes captured)Address Resolution Protocol (request) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode request (0x0001) Sender MAC address DellComp_5e40b9 (00065b5e40b9) Sender IP address 12811924581 (12811924581) Target MAC address 000000_000000 (000000000000) Target IP address 128119245254 (128119245254)
Frame 204 (60 bytes on wire 60 bytes captured)Address Resolution Protocol (reply) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode reply (0x0002) Sender MAC address DigitalE_00e80b (aa000400e80b) Sender IP address 128119245254 (128119245254) Target MAC address DellComp_5e40b9 (00065b5e40b9) Target IP address 12811924581 (12811924581)
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
ARP trace
Frame 203 (42 bytes on wire 42 bytes captured)Address Resolution Protocol (request) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode request (0x0001) Sender MAC address DellComp_5e40b9 (00065b5e40b9) Sender IP address 12811924581 (12811924581) Target MAC address 000000_000000 (000000000000) Target IP address 128119245254 (128119245254)
Frame 204 (60 bytes on wire 60 bytes captured)Address Resolution Protocol (reply) Hardware type Ethernet (0x0001) Protocol type IP (0x0800) Opcode reply (0x0002) Sender MAC address DigitalE_00e80b (aa000400e80b) Sender IP address 128119245254 (128119245254) Target MAC address DellComp_5e40b9 (00065b5e40b9) Target IP address 12811924581 (12811924581)
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Routing to another LAN
walkthrough send datagram from A to B via R assume A knows B IP address
Two ARP tables in router R one for each IP network (LAN)
A
RB
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
A creates datagram with source A destination B A uses ARP to get Rrsquos MAC address for 111111111110 A creates link-layer frame with Rs MAC address as dest
frame contains A-to-B IP datagram Arsquos adapter sends frame Rrsquos adapter receives frame R removes IP datagram from Ethernet frame sees its
destined to B R uses ARP to get Brsquos MAC address R creates frame containing A-to-B IP datagram sends to
B
A
RB
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Medium access control
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-18
Multiple Access Links and Protocols
Two types of ldquolinksrdquo point-to-point
PPP for dial-up access point-to-point link between Ethernet switch and host
broadcast (shared wire or medium) old-fashioned Ethernet upstream HFC 80211 wireless LAN
shared wire (eg cabled Ethernet)
shared RF (eg 80211 WiFi)
shared RF(satellite)
humans at acocktail party
(shared air acoustical)
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-19
Multiple Access protocols
single shared broadcast channel two or more simultaneous transmissions by nodes
interference collision if node receives two or more signals at the same
time
multiple access protocol distributed algorithm that determines how nodes
share channel ie determine when node can transmit
communication about channel sharing must use channel itself no out-of-band channel for coordination
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-20
Ideal Multiple Access Protocol
Broadcast channel of rate R bps1 when one node wants to transmit it can send at
rate R2 when M nodes want to transmit each can send at
average rate RM3 fully decentralized
no special node to coordinate transmissions no synchronization of clocks slots
4 simple
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-21
MAC Protocols a taxonomy
Three broad classes Channel Partitioning
divide channel into smaller ldquopiecesrdquo (time slots frequency code)
allocate piece to node for exclusive use
Random Access channel not divided allow collisions ldquorecoverrdquo from collisions
ldquoTaking turnsrdquo nodes take turns but nodes with more to send can
take longer turns
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-22
Channel Partitioning MAC protocols TDMA
TDMA time division multiple access access to channel in rounds each station gets fixed length slot (length = pkt
trans time) in each round unused slots go idle example 6-station LAN 134 have pkt slots
256 idle
1 3 4 1 3 4
6-slotframe
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-23
Channel Partitioning MAC protocols FDMA
FDMA frequency division multiple access
channel spectrum divided into frequency bands each station assigned fixed frequency band unused transmission time in frequency bands go
idle example 6-station LAN 134 have pkt
frequency bands 256 idle fr
eque
ncy
band
s time
FDM cable
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-24
Random Access Protocols
When node has packet to send transmit at full channel data rate R no a priori coordination among nodes
two or more transmitting nodes ldquocollisionrdquo random access MAC protocol specifies
how to detect collisions how to recover from collisions (eg via delayed
retransmissions)
Examples of random access MAC protocols slotted ALOHA ALOHA CSMA CSMACD CSMACA
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-25
Slotted ALOHA
Assumptions all frames same size time divided into
equal size slots (time to transmit 1 frame)
nodes start to transmit only slot beginning
nodes are synchronized
if 2 or more nodes transmit in slot all nodes detect collision
Operation when node obtains fresh
frame transmits in next slot if no collision node
can send new frame in next slot
if collision node retransmits frame in each subsequent slot with prob p until success
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-26
Slotted ALOHA
Pros single active node can
continuously transmit at full rate of channel
highly decentralized only slots in nodes need to be in sync
simple
Cons collisions wasting
slots idle slots nodes may be able to
detect collision in less than time to transmit packet
clock synchronization
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-27
Slotted Aloha efficiency
suppose N nodes with many frames to send each transmits in slot with probability p
prob that given node has success in a slot = p(1-p)N-1
prob that any node has a success = Np(1-p)N-1
max efficiency find p that maximizes Np(1-p)N-1
for many nodes take limit of Np(1-p)N-1
as N goes to infinity gives
Max efficiency = 1e = 37
Efficiency long-run fraction of successful slots (many nodes all with many frames to send)
At best channelused for useful transmissions 37of time
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-28
Pure (unslotted) ALOHA
unslotted Aloha simpler no synchronization when frame first arrives
transmit immediately
collision probability increases frame sent at t0 collides with other frames sent in [t0-
1t0+1]
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-29
Pure Aloha efficiency
P(success by given node) = P(node transmits)
P(no other node transmits in [t0-
1t0]
P(no other node transmits in [t0t0+1]
= p (1-p)N-1 (1-p)N-1
= p (1-p)2(N-1)
hellip choosing optimum p and then letting n -gt infty
= 1(2e) = 18
even worse than slotted Aloha
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-30
CSMA (Carrier Sense Multiple Access)
CSMA listen before transmitIf channel sensed idle transmit entire frame If channel sensed busy defer transmission
human analogy donrsquot interrupt others
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-31
CSMA collisions
collisions can still occurpropagation delay means two nodes may not heareach otherrsquos transmission
collisionentire packet transmission time wasted
spatial layout of nodes
noterole of distance amp propagation delay in determining collision probability
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-32
CSMACD (Collision Detection)
CSMACD carrier sensing deferral as in CSMA collisions detected within short time colliding transmissions aborted reducing
channel wastage collision detection
easy in wired LANs measure signal strengths compare transmitted received signals
difficult in wireless LANs received signal strength overwhelmed by local transmission strength
human analogy the polite conversationalist
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-33
CSMACD collision detection
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-34
ldquoTaking Turnsrdquo MAC protocols
channel partitioning MAC protocols share channel efficiently and fairly at high load inefficient at low load delay in channel access
1N bandwidth allocated even if only 1 active node
Random access MAC protocols efficient at low load single node can fully utilize
channel high load collision overhead
ldquotaking turnsrdquo protocolslook for best of both worlds
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-35
ldquoTaking Turnsrdquo MAC protocols
Polling master node
ldquoinvitesrdquo slave nodes to transmit in turn
typically used with ldquodumbrdquo slave devices
concerns polling overhead latency single point of
failure (master)
master
slaves
poll
data
data
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-36
ldquoTaking Turnsrdquo MAC protocols
Token passing control token passed
from one node to next sequentially
token message concerns
token overhead latency single point of failure
(token)
T
data
(nothingto send)
T
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-37
Summary of MAC protocols
channel partitioning by time frequency or code Time Division Frequency Division
random access (dynamic) ALOHA S-ALOHA CSMA CSMACD carrier sensing easy in some technologies (wire)
hard in others (wireless) CSMACD used in Ethernet CSMACA used in 80211
taking turns polling from central site token passing Bluetooth FDDI IBM Token Ring
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Ethernet hubs switches routers
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Ethernet
ldquodominantrdquo wired LAN technology cheap - $20 for 1000Mbs first widely used LAN technology Simpler cheaper than token LANs and ATM Kept up with speed race 10 Mbps ndash 10 Gbps
Metcalfersquos Ethernetsketch
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
5 DataLink Layer 5-40
Star topology bus topology popular through mid 90s
all nodes in same collision domain (can collide with each other)
today star topology prevails active switch in center each ldquospokerdquo runs a (separate) Ethernet protocol
(nodes do not collide with each other)
switch
bus coaxial cable star
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Ethernet Frame Structure
Sending adapter encapsulates IP datagram (or other network layer protocol packet) in Ethernet frame
Preamble 7 bytes with pattern 10101010 followed by one
byte with pattern 10101011 used to synchronize receiver sender clock
rates
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Unreliable connectionless service
Connectionless No handshaking between sending and receiving adapter
Unreliable receiving adapter doesnrsquot send acks or nacks to sending adapter stream of datagrams passed to network layer
can have gaps gaps will be filled if app is using TCP otherwise app will see the gaps
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Ethernet uses CSMACD
No slots adapter doesnrsquot
transmit if it senses that some other adapter is transmitting that is carrier sense
transmitting adapter aborts when it senses that another adapter is transmitting that is collision detection
Before attempting a retransmission adapter waits a random time that is random access
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Ethernet CSMACD algorithm
1 Adaptor receives datagram from net layer amp creates frame
2 If adapter senses channel idle it starts to transmit frame If it senses channel busy waits until channel idle and then transmits
3 If adapter transmits entire frame without detecting another transmission the adapter is done with frame
4 If adapter detects another transmission while transmitting aborts and sends jam signal
5 After aborting adapter enters exponential backoff after the mth collision adapter chooses a K at random from 012hellip2m-1 Adapter waits K512 bit times and returns to Step 2
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Ethernetrsquos CSMACD (more)
Jam Signal make sure all other transmitters are aware of collision 48 bits
Bit time 1 microsec for 10 Mbps Ethernet for K=1023 wait time is about 50 msec
Exponential Backoff Goal adapt retransmission
attempts to estimated current load heavy load random wait
will be longer first collision choose K
from 01 delay is K 512 bit transmission times
after second collision choose K from 0123hellip
after ten collisions choose K from 01234hellip1023
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Interconnecting with hubs Backbone hub interconnects LAN segments Extends max distance between nodes But individual segment collision domains become
one large collision domain Canrsquot interconnect 10BaseT amp 100BaseT
hub
hubhub
hub
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Switch
Link layer device stores and forwards Ethernet frames examines frame header and selectively
forwards frame based on MAC dest address when frame is to be forwarded on segment
uses CSMACD to access segment transparent
hosts are unaware of presence of switches plug-and-play self-learning
switches do not need to be configured
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Forwarding
bull How do determine onto which LAN segment to forward framebull Looks like a routing problem
hub
hubhub
switch1
2 3
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Self learning
A switch has a switch table entry in switch table
(MAC Address Interface Time Stamp) stale entries in table dropped (TTL can be 60
min) switch learns which hosts can be reached through
which interfaces when frame received switch ldquolearnsrdquo location
of sender incoming LAN segment records senderlocation pair in switch table
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
FilteringForwarding
When switch receives a frame
index switch table using MAC dest addressif entry found for destination
then if dest on segment from which frame arrived
then drop the frame else forward the frame on interface indicated else flood
forward on all but the interface on which the frame arrived
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Switch example
Suppose C sends frame to D
Switch receives frame from from C notes in bridge table that C is on interface
1 because D is not in table switch forwards
frame into interfaces 2 and 3 frame received by D
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEG
1123
12 3
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Switch example
Suppose D replies back with frame to C
Switch receives frame from from D notes in bridge table that D is on interface 2 because C is in table switch forwards frame
only to interface 1 frame received by C
hub
hub hub
switch
A
B CD
EF
G H
I
address interface
ABEGC
11231
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Switch traffic isolation switch installation breaks subnet into LAN segments switch filters packets
same-LAN-segment frames not usually forwarded onto other LAN segments
segments become separate collision domains
hub hub hub
switch
collision domain collision domain
collision domain
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Switches dedicated access
Switch with many interfaces
Hosts have direct connection to switch
No collisions full duplex
Switching A-to-Arsquo and B-to-Brsquo simultaneously no collisions
switch
A
Arsquo
B
Brsquo
C
Crsquo
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
More on Switches
cut-through switching frame forwarded from input to output port without first collecting entire frameSlight reduction in latencyWas a big deal in the days of
10Mbits Q could we do this on routers
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Typical institutional network
hub
hubhub
switch
to externalnetwork
router
IP subnet
mail server
web server
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Switches vs Routers both store-and-forward devices
routers network layer devices (examine network layer headers) switches are link layer devices
routers maintain routing tables implement routing algorithms
switches maintain switch tables implement filtering learning algorithms
Switch
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Summary comparison
hubs routers switches
traf f ic isolation
no yes yes
plug amp play yes no yes
optimal routing
no yes no
cut through
yes no yes
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Bridging LANs
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Self-learning =gt Loops
Assume X sends a frame to Y
B1 repeats it first and B2 receives it on interface 2
What does B2 do
B1 B2
X
Y
1
2
1
2
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Self-learning =gt Broadcast storms
Frames may reproduce with more bridges
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Spanning tree protocol
Spanning tree protocol prevents loops and broadcast stormsAvoids some links to prevent
loopsstormsSpanning tree = acyclic subgraph
covering all vertices
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Protocol requirements
Should allow redundant bridges connected ldquocarelesslyrdquo
Self-configuring or plug-and-play Small memory usage Constant message overhead in each
LAN Quick stabilization to loop-freeness Assume connectionless service ie
messages may be lost
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Algorithm
Elect a root Root = switch with the smallest ID
Nodes periodically send HELLO message Transmitting node ID ID of bridge assumed to be root Length of best known path to root
Node = bridge or LAN segment Edge = from a switch interface to LAN or
interface in adjacent switch
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Algorithm
Goal to identify if an interface is on shortest path to root
Start each switch X thinks it is root Announces (XX0)
Distributed asynchronous step Update root to node with minimum ID Announce 1 + length of minimum distance
heard from neighbors to that root Accept neighbor closer to root as ldquodesignatedrdquo
for that LAN segment and stop sending it HELLOs
Or designate self
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Example
4 thinks it is root Sends (440) to 2 and 7
4 receives (220) from 2 Thinks 2 is root and just
one hop away 4 receives (271) from 7
Prefers 2 over 7 to reach root
Removes link 4-7 from tree
Designates 2 and stops sending it messages
1
2
3
4
5
67
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Tolerating failures
Soft state approachReceived messages have an expiry time
MAX_AGE If no new messages received after expiry
nodes will try to take over as rootMAX_AGE gt MaxPropTime (Why)Re-convergece time after failure =
MAX_AGE + 2MaxPropTime (Why)
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
State machine
PRE_BACKUP_DELAY to prevent transient partitions gt 2MAX_AGE Why
PRE_FORWARD_DELAY to prevent transient loops 3MAX_AGE Why
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Virtual LANs
Switched LANs have limited scalabilityLinear scaling behavior of spanning treeBroadcast packets sent to all nodesMovement across domains a problem
Solution virtual LANs (VLANs)LANs partitioned using colorsEg red packets forwarded only to red
LANs Q VLANs vs subnets difference
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
Discussion questions
The plug-and-play property of switches implies less management overhead Why are routers not plug-and-play
Each interface has a MAC address and an IP address Why canrsquot we just route over MAC addresses
IP addresses conflate location and names Implications on multihoming Implications on mobility
