CANARIE
“CA*net 4Customer Empowered Networking”
http://www.canarie.ca
http://www.canet3.net
[email protected]: +1.613.785.0426
Mission: To facilitate the development of Canada’s communications infrastructure and stimulate next generation products, applications and services
Canadian equivalent to Internet 2 and NGI private-sector led, not-for-profit consortium consortium formed 1993 federal funding of $300m (1993-99) total project costs estimated over $600 M currently over 140 members; 21 Board members
CANARIE Inc
The Context Telecom market trapped in a nuclear winter with no end in
sight Bankrupt CLECs Failed wireless companies Troubled ILECs Huge over capacity and financial bubble
The promise of de-regulation and competition are fading Return to a healthy and profitable industry is a long way
off But despite all the gloom - there is a glimmer of hope in a
new business model that may return the industry to profitability…..
Customer Owned Networks The customer owns the infrastructure (dark fiber, switches and
wavelengths) while the carrier provides the service and network management
Relieves the carrier of huge capital cost of infrastructure and gives customer greater flexibility in choice of service provider and control of the network
Very similar parallel to evolution of computer industry from the centrally managed time share of the 1960s to the customer owned mini-computer of the 70s and the PC of the 80s
Today telecom is largely a service industry much like time share computing of the 60s
Asset based telecom puts customer in control and ownership of the network
Asset based telecom started with the same people who brought you the Internet – our universities and research centers
Customer owned networks Most Canadian universities and research centers own and
control local loop dark fiber Many regional networks also have their own fiber networks
BCnet, Netera, RISQ, ORION, ACORN On a national basis CA*net 4 in Atlantic Canada
CANARIE will own the fiber and equipment and operate infrastructure as a condominium
Other organizations can buy into condominium Group Telecom (Chapter 11 CLEC) will operate and manage the
equipment under contract to the condominium association CANARIE gets 16 wavelengths, carrier gets 16 wavelengths to sell
managed services or to other condominium owners If GT goes bankrupt, condominium association can find another
company to manage the network
OC-12
MAN
Scale 100 Km
RISQ Fiber Network
Dark fiberLeased bandw.
250 Km
450
Km
Network
OC-48
Level 3 provides dark fiber for California Research Network
BOEING BUILDS PRIVATE NATIONWIDE OPTICAL NETWORK
Boeing awarded an estimated US$20 million contract to Nortel Networks to build a private nationwide optical network based on DWDM and next generation SONET. Plans call for the deployment of OPTera Long Haul 1600 Optical Line Systems and OPTera Metro 3500 Multiservice Platform in multiple cities across the country. http://www.nortelnetworks.com/corporate/news/newsreleases/2002c/07_23_02_boeing.html
Circuit Switched Networks!! Recent research from Dr. Malathi Veeraraghavan – NY Polytechnic With hardware signaling, even at low RT propagation delay and long-
path environments, Circuit Switched does better 3.5 times lower delay at a 50ms prop. delay at 1Gbps for a 20MB
file Zing: a transport protocol that uses an “end-to-end” circuit in
conjunction with TCP/IP path Circuit switching not inherently bad
The telco business/engineering model has given CS a bad name Can we take attributes and benefits of connectionless network like
Internet, particularly end 2 end principle, and apply to CS networks?
The Problem with CS Current optical switched circuits are “edge to edge” within a
single carrier cloud Any changes to circuit in terms of bandwidth or topology requires release
of current circuit and establishment of new circuit Customer cannot make autonomous topology or bandwidth changes or
cross connect to another circuit within the carrier cloud
No inter-domain optical routing protocol Customer cannot set up an end to end wavelength across multiple
domains No optical services for end2end light path across the campus/enterprise
and across the carrier cloud
All current optical services are based on a client –server model No ability to exchange wavelengths and services on a peer to peer basis
Approaches to Circuit Switching Two solutions:
Flow switching (e.g Ipsilon, ATM) Application switching
Flow switching has major limitations: Out of packet sequence problems on initiation of flow Detecting beginning and ending of long flows requires a lot of packet
sniffing, timers and state Difficult to setup across interdomain networks
Application switching Allows application to direct flow to dedicated channel e.g GridFTP Allows OS bypass and large MTU and non TCP for greater efficiency Allows web service interface for signal and setup of dedicated path
The Future -CA*net 4? Funded by Gov’t of Canada for $110m – now fully operational A network of point to point condominium wavelengths Universities and researcher own and control their own lightpaths or wavelengths and
associated cross connects on each switch All lightpaths terminate at switches where condominium owner can manage their own
portion of the switch add/drop STS channel or lightpaths cross connect to another condominium owner’s STS channels or wavelengths
Web service architecture for management of optical networks Plan to use OGSA for optical management
Owners of wavelengths determine topology and routing of their particular light paths Condominium owner can recursively sub partition their wavelengths and give ownership
to other entities Wavelengths become objects complete with polymorphism, inheritance, classes, etc –
Object Oriented Networking
CA*net 4 Architecture
Calgary ReginaWinnipeg
OttawaMontreal
Toronto
Halifax
St. John’s
Fredericton
Charlottetown
ChicagoSeattleNew York
CANARIEGigaPOP
ORAN DWDM
Carrier DWDM
Thunder Bay
CA*net 4 node)
Possible future CA*net 4 node
Quebec
Windsor
Edmonton
Saskatoon
Victoria
Vancouver
Boston
Initial Version of Mini-IXExternal Proxy Server
CA*net 4 Proxy Server
Standard CLI or TL1 interface
Customer A and sub- partition
Customer B
Customer C
OC192 EastboundOC192 Westbound
X
X
X
OSPF
OBGP
OBGP
Customer A signaling plane
Subtended GbE to local GigaPOP
Customer B signaling plane
Grooming agents
Switch Agents
Customer C signaling plane
Signal Control Plane Agents
X X
Customer A Proxy Server
Example
LPO I/F
LOS Server
LPO LPO LPO
LPO I/F
LOS Server
LPO LPO
LPO I/F
LOS Server
LPO LPO
LPO I/F
LOS Server
LOS Server
LPORegistry
End2End Lightpath Creation Agent
AS1 AS2 AS3 AS4
x.x.x.1
Local Border Router
LPO I/F
Advertises LPO between AS2 and AS4
Destination
WSIL pointers, UDDI or IRR
LPO I/F
AS5
1 2
3
4
5
802.1 p/q VLANServer
Campus VLAN to OBGP
End2End Lightpath Creation Agent
LOS Server
LPO
LPO I/F
LPO I/F
LPO
Inspection.wsil
VLAN
3
4
1
2
2
Campus Border Router
1.Border Router advertises VLAN lightpath and external lightpath to GigaPOP2. End user creates lightpath object linking 2 objects3. End user (may) advertise newly created lightpath to other entities
End user
OBGP Proposed new protocol to allow customer owned wavelengths
to interconnect to each other at an optical switch Optical switch is in effect a mini-IX Use establishment of BGP or peers for process to establish
light path cross connects Allows network admin to maintain routing policy on cross
connect even though there may be a cut through path Traditional BGP gives no indication of route congestion or
QoS, but with DWDM wave lengths edge router will have a simple QoS path of guaranteed bandwidth
OBGP can use “optical” AS path to concatenate wavelengths across multiple AS to have continuous QoS path
The future?Self Organizing Networks- RPON
Passive OpticalSplitter
Aggregator
ISP
NeighborhoodNode
Switch
Customer Controlled or Owned Fiber
Active laser at customer premises
Conclusion The customer owns the infrastructure (dark fiber,
switches and wavelengths) while the carrier provides the service and network management
Relieves the carrier of huge capital cost of infrastructure and gives customer greater flexibility in choice of service provider and control of the network
Universities and research networks are leading the way Significant opportunity for Taiwan telecom and optical
industry to develop the technologies required for customer owned network