Introduction..........................................................................................................................2Why This Book ?......................................................................................................... 2Ethernet in the Market..................................................................................................3Ethernet Everywhere....................................................................................................8A Case in Point - Mobile Ethernet Back Hauling...................................................... 14Evolution of Ethernet................................................................................................. 19
ALOHA Network...............................................................................................19Xerox PARC Years............................................................................................21After Xerox........................................................................................................ 22
Introduction
Why This Book ?The topic of this book is Carrier Ethernet. What is Carrier Ethernet ? Carrier
Ethernet is the Ethernet technology deployed on the scale of MEN and WAN with
characteristics for the traditional carrier networks such as SONET/SDH1.
Why this topic is the subject of the book? Beginning from its humble origins in 70th
of XX century in the labs of Xerox Palo Alto Research Center (PARC) low bandwidth,
low quality or no quality, local networking technology designed to connect few
workstations over a piece of coaxial cable, in the last decade of XX century developed
into the most ubiquitous networking technology beating such sure winners like ATM, FR,
FDDI, and even SONET/SDH, not talking about TDM technologies (DS1/T1/T3s) - the
workhorse of early telecommunication networks and many others2. Most of these
networking technologies are already the name of the past, like FOTRAN or CPM OS
became for computing technology, and some are completely forgotten. What is and will
be around is Ethernet3 in LAN and Carrier Ethernet in MAN and WAN.
Thus, Ethernet as a networking technology, already is and will be for years to come
(we do not know yet what will be the next technology ) The Technology of the networks.
1 We assume that the reader has some level of knowledge of the networking technology and in the Introduction we donot explain the meaning of abbreviations. All of these terms will be explained in details in subsequent parts of the book.2 Dell’Oro Ethernet Switch Forecast 2014-2018 indicates that the shipment of the Token Ring and LAN ATM portsdies out around 2003-2004. http://www.ethernetalliance.org/wp-content/uploads/2013/12/OFC_roadmap.pdf.Retrieved 12.15.2014.3 The statements to this fact are plenty. For example, this statements comes from 2005 from the book on Cisco LANSwitching” Unlike Token Ring and FDDI, which for the most part are defunct, Ethernet technology is alive and verywell” Barnes David, Basir Sankadar. Cisco LAN switching Fundamentals. CiscoPress, Indianapolis, 2005. Already in1998, in 25th Anniversary of Ethernet invention Ethernet was becoming the force in the networking- “Today Ethernet ismore than just another type of LAN; it is de facto standard for local area networking hardware, with 200 million nodesinstalled worldwide.”Breyer Robert and Sean Ripley. Switching, Fast, and Gigabit Ethernet. 3rd ed. MacmillanTechnical Publishing. USA. 1999.
Ethernet in the MarketA look at market data brings out the Ethernet success story even more clearly.
Vertical Systems in the article published in Carrier Ethernet News4 in 2013 predicts that
Carrier Ethernet services by 2017 will reach $49.4 billions in revenue. This is 5-fold
growth from nearly $10 billions in 2007.
Recent data from the Ethernet switch market indicate the substantial growth of 11%
in 2Q 2014 to $5.4 billions5 ( see Figure 1.1). In 2013 the same period showed 20%
growth.
Figure 1.1 Ethernet Port shipment in 2013-214.
In the same period the shipment of 100G ports grew over 50%. The number of
shipped ports in 1G and 10 G in copper and fiber tells the similar story6. These trends are
documented by numbers in Tables 1.1 and 1.2 and Figures 1.2 and 1.3.
4 Global Ethernet Services Market Headed For Nearly $50 Billion By 2017. Vertical Systems, May , 2013. Accessedon 12.12.2014 athttp://www.carrierethernetnews.com/articles/634000/global-Ethernet-services-market-headed-for-nearly-/5 http://www.cablinginstall.com/articles/2014/09/infonetics-ethernet-switching-bounceback.html. Retrieved12.12.2014.6 Based on the Dell’Oro Ethernet Switch Forecast 2014-2018;http://www.ethernetalliance.org/wp-content/uploads/2013/12/OFC_roadmap.pdf. Retrieved 12.15.2014.
2012 2013 2014 2015 2016 2017 2018Copper 208,271 240,577 266,943 279,474 286,858 292,207 293,873Fiber 14,322 13,495 13,014 11,877 10,407 8,765 7,040Total 222,593 254,072 279,958 291,351 297,265 299,973 300,913% Copper 94% 95% 95% 96% 97% 97% 98%
Table 1.1 The Number of Shipped 1GE Ports (in 000s) from 2012 to 2018 (predicted).
Figure 1.2 The Number of Shipped 1GE Ports from 2012 to 2018 (predicted).
2012 2013 2014 2015 2016 2017 2018Copper 6,846 13,018 21,109 31,210 41,486 50,019 58,758Fiber 8,923 9,341 9,113 11,558 13,271 15,120 14,716Total 15,796 22,359 30,223 42,768 54,757 65,138 73,475% Copper 43% 58% 70% 73% 76% 77% 80%
Table 1.2 The Number of Shipped 10GE Ports (in 000s) from 2012 to 2018 (predicted).
Figure 1.3 The Number of Shipped 10 GE Ports from 2012 to 2018 (predicted).
While it is expected that the shipment of 1GE ports will stabilize towards 2018 (Figure
1.2, Table 1.1) , the shipment of 10 GE ports will continually grow.
In January 2014 Cable Installation & Maintenance (Figure 1.4) announced7:
“ 10 Gigabit Ethernet is finally on the verge of becoming the most popular data
center switch port connection, after a long and sometimes rocky adoption
curve,” said Seamus Crehan, president of Crehan Research, when announcing
the findings of the firm’s latest report. He followed up that comment with a rosy
outlook for 40GbE: “As 40GbE starts to ramp, we are still forecasting its
adoption curve to look much better than that of 10GbE. This is already
evidenced by the fact that recent data center switch introductions are really
pushing the envelope on 40GbE port densities and economics.”
7 Researcher: 10G will account for more than half of Ethernet data center switch ports shipped this year.http://www.cablinginstall.com/articles/2014/01/crehan-10g-2014.html. Retrieved 12.14.2014
Table 1.4 Trends in 10 GE, FC and FDR shipments for Data Centers8.
Thus, the future for the Ethernet industry looks bright. Looking into the 100GE
market it seems even brighter. That is what the Infonetics analysts predicts9:
“Overall network port shipments and revenue are on a steady upward path
as buyers shift to higher bandwidth, but the real action is in high-speed (10G+)
port shipments, which we expect to increase almost ten-fold by 2017,” notes
Matthias Machowinski, directing analyst for enterprise networks and video at
Infonetics Research. Andrew Schmitt, principal analyst for optical at Infonetics,
adds: “For optical ports, 10G will remain the highest-volume speed, but 100G
represents the area of the most dramatic growth. Service providers have
indicated to us that by 2016, the majority of spending in long-haul networks
will be on 100G.”
Figure 1.5 shows the global trends in Carrier Ethernet published in 2012 by Ovum
and provided by Metro Ethernet Forum10. Even in the ‘pessimistic’ scenarios the Ethernet
8 FDR is a technology used in InfiniBand (IB) (intra, inter computer high band, low latency channel) connections; anevolution step in SDR DDR, QDR,FDR, EDR, HDR development path.9 http://www.infonetics.com/pr/2013/1H13-Networking-Ports-Market-Highlights.asp. Retrieved 12.14.2014.10 An Overview of the Work of MEF. http://metroethernetforum.org/carrier-ethernet/presentations. Retrieved
market will record the substantial growth. In fact as we see the market in 2014 the small
dip in the demand seen on the pessimistic line, was recorded. But the market bounced
back.
Figure 1.5 Global Carrier Ethernet Trends as published by Ovum in 2012.
In 2014 MEF published Vertical Systems Group statistics showing that the total
worldwide bandwidth purchased for the Ethernet services exceeded the legacy; the
cross-over point was somewhere in 2012, as seen on Figure 1.6.
Figure 1.6 Legacy and Ethernet Data Traffic.
Of course, whatever the past and present times are telling us, it is only the future will
show whether the Ethernet expansion will continue for several years to come. However,
there is no obvious contender for this technology on the horizon.
12.15.2014.
Ethernet EverywhereTo really comprehend the evolution that Ethernet technology made from its invention
in Palo Alto Xerox Labs to Carrier networks one must look at the scale of the networks
the Ethernet technology is being deployed as seen on Figure 1.7. In its inception it was to
operate as LAN technology on the scale of an office, office
Figure 1.7 Types of Networks.
building or a campus. As Ethernet benefits were realized, or shortcoming of other
technologies became apparent, Ethernet moved into Access and MAN networks11. And
eventually into the WAN transport networks extending on the scale of continents and
beyond. Each move from LAN to MAN to WAN required news features to be supported
by the technology and posed obviously new technical challenges. It is not only the jump
in the bandwidth from 2.96 Mbps to 100 G and 400 on the horizon, but also the change in
transport media from coaxial cable to fiber optics, the change in the control and
management planes. One may say that the change of scope of services required
reinvention of technology. Usually, technologies were designed for the specific
application/area/service. The Ethernet turned out to be so flexible that was able to evolve
into new areas and services as needs demanded.
11 The detailed review of LAN/MAN and WAN networking technologies will be provided in the next chapter.
The requirements for each of the network types shown in Table 1.3.
Table 1.3 LAN, MAN and WAN Characteristics
The LAN networks are deployed on the scale of the campus, office or and office
complex. They usually are under a single management and provide a limited range of
services. It connects the end user equipment with services, devices such as printers,
servers, storage and provides the access to the Internet. The resiliency is not of the utmost
priority ( with exception of the core equipment) with 3 9s being a usual availability target.
SLAs are not deployed. However, the internal performance targets ( SLT- Service Level
Targets) may be defined. The expected bandwidth ranges from 100Mbps to 1000 Mbps or
more.
The Access is usually referred to as the first or last mile and extends from the
customer premises to the first provider access point ( edge switch or router)- usually not
LAN MAN WANGeographic Coverage Office or office complex City or city and suburbs State, county, country,
continentsScale Tens to thousands end
points ( PCs, Servers,Printers), switches, NIDs
Multiple access nodes,multiple aggregationswitches, redundant paths
Multiple access nodes,highly redundantarchitecture,
Technology Usually SingleTechnology
Must support multipleaccess technologies
Few interconnectiontechnologies
Interfaces UNI to MAN UNI to MAN, ENNI toWAN
ENNI
Services Usually few servicesdesigned for the specificenterprise
Several standardizedservices with severalfeatures supporting manypotential customer types,high granularity of services
Fewer services, smallgranularity, highbandwidth
Management A single company May have MultipleProviders
Few Providers
Interface Bandwidth 100 to 1000 Mbps 10 to 100 Gbps 10 to 100 GbpsTraffic CarryingCapacity
several Gbps Few Tbps Few Tbps
Resiliency, availability Not critical, 3 ninesavailability or more
Critical, required 4 or 5nines availability
Critical required > 4nines availability
QoS Not often implemented Up to 4 ( or more) QoSclasses
Usually low (2) numberof QoS classes
SLA Usually not Specified ormonitored, may haveSLSs
Strict SLAs tied to the typeof service
Strict SLAs tied to thetype of service
more than few miles. It is part of the MAN or METRO network with the sole purpose to
connect LAN locations across the city or the country and provide the access to services
and facilities not available locally. Access networking technologies are describes in more
details in the subsequent chapters.
MAN networks link business locations, government service centers, schools, and
provide the support for retail services across the city. The radius of METRO/MAN
networks may extend to 100 - 120 miles and is limited by the requirements of service or
technology itself. For example, limiting factor may be the maximum one-way latency in
the overall transport budget for the service or the number of hops. MAN networks may be
composed of sub-networks under different management. They have to offer a wide range
of services to suit the variety of customers. Thus, MAN networks have strict SLAs
imposed, with very strict target metrics of delay, jitter and packet loss. The resiliency
required by MAN networks are at the minimum 4 9s. The networks are composed mostly
from 10 Gig trunks with the total capacity going even to few Tbps for very large
metropolitan areas.
The connectivity between cities and continents is implemented with WAN networks
that extend over 1000s of miles. They provide national and continental networking
infrastructure. They usually have 10 Gbps links, very high availability ( 4 or 5 9s), lower
than MAN number of QoS classes and strict performance targets for three basic metrics:
delay, jitter (delay variation), packet loss.
One may ask what is so special in the Ethernet technology that it eliminated most of
the other networking solutions in the LAN, MAN, and WAN networks? Factors for
success of Ethernet are several12. They may be grouped in four major areas:
Convergence technology support
Cost per port and bit
Bandwidth and bandwidth efficiency
12 25 years after the invention of Ethernet the domination of the LAN market by Ethernet was obvious and at thattime the reasons for this were states as “.. Ethernet was continuously reinvented itself to keep up with rapidly changingmarkets requirements, and ... New Ethernet technology have evolved the price/performance curve without makingobsolete past iterations”. Breyer Robert and Sean Ripley. Switching, Fast, and Gigabit Ethernet. 3rd ed. MacmillanTechnical Publishing. USA. 1999.
Provisioning and Operations factors.
Convergence technology -This feature of the Ethernet technology has the biggest
impact in MAN networks. Ethernet provides a common networking solution for many
networking technologies in so called first mile segments or an access. Segment of the
network. The First mile or last mile is a network part that connects the customer location
with the MAN transport network. Figure 1.8 illustrates the technical challenge facing the
first mile engineering. The first mile may be engineered using the copper line,
SONET/SDH loop, TDM T1/T3 line, HFC access, GPON, direct fiber, WDM fiber,
ADSL, or WiMAX (Table 1.4)
With multiple access technologies Ethernet allows the platform in the MAN support
the traffic from any access to any access. Thus, one type of network can support any of
the existing access methods. With the convergence of technology came the convergence
of services; Ethernet support, which was only wished for in the last decade of the past
Century13, of the converged data/voice/ video traffic over packet networks became
realized and spurred the explosive growth of Ethernet technology in the second half of the
first decade of XXI century.
13 In 1999 the following statement was made “ ... The biggest challenge for Ethernet still lies ahead, with theconvergence of data, voice, and video communication. The consolidation of the traditional data, voice, and videotelecommunication services will result in the growth of the packet-switch-based-networks and Ethernet is very wellpositioned to carry the IP traffic of the integrated data/voice/video networks of the future”. Breyer Robert and SeanRipley. Switching, Fast, and Gigabit Ethernet. 3rd ed. Macmillan Technical Publishing. USA. 1999. Few predictionswere as true as this one.
Figure 1.8 Multiple Access Technologies in MAN14
Type of Access Technology in the 1stMile
Direct Fiber
WDM Fiber
Cable COAX
E/G/PON
Bonded T1/E1 TDM
DS1/DS3
SONET/SDH
Packet Wireless
WiMax
xDSL
Table 1.4 Types of Access Technology
Cost per bit/port - Ethernet technology offers lower cost per bit and port than other
LAN technologies. Simply with the Ethernet the cost of the service is not growing
linearly. At one point efficiency of scale ( the number of shipped ports lower the price per
port) helped Ethernet to beat networks such as Token Ring, offering the simply cheaper
solutions. The Case in Point section describes the story of the explosion of Ethernet
services for the LTE. The story that in fact mimic the success of Ethernet in the market.
The ascend of Ethernet in the wireless back hauling was exactly based on two factors
- low cost per bit and non-linear ( slower) growth of cost per bit in the function of
capacity. Simply, the older technology (TDM) could need keep up with the demand ( of
course the other factors such as maturity of Ethernet, ubiquitous presence and Carrier
features support were also the critical factor in the LTE Ethernet evolution). This is a
14 Extending Ethernet into the First Mile. MEF Reference Presentation, 2011.
statement from RCR Wireless from 201315:
“For a variety of reasons, Layer 2 Ethernet has emerged as the overwhelming
choice for data-centric LTE backhaul connectivity. Whether delivered over
fiber or packet (microwave/millimeter) radio, backhaul providers rely on
Ethernet to scale the network quickly and achieve the lowest cost-per-bit while
increasing bandwidth to meet growing user demand. Equally important, the
inherent flexibility of Ethernet helps backhaul providers avoid costly
over-provisioning of the network, even as mobile network operators demand
multiple classes of service across it. Based on these advantages, Infonetics
predicts that Ethernet will account for more than 80% of all backhaul services
revenue by 2015.”
As with all predictions they are outdated sooner than their due date arrives. In few
years this piece of futuristic thinking will be history, but in its years such views had
driven the technology.
Bandwidth and Bandwidth Efficiency - Ethernet technology offers bandwidth (port
capacity) from 10Mbps to 100 Mbps, 1G, 10 G and now 100G (400G on the horizon)
over the same network. And within each port Ethernet services may be crafted to offer a
finely granulated services of different QoS specifications including CIR, EIR, CBS and
EBS parameters. The Ethernet frame is more efficient than the network technologies such
as SONET or ATM using the fixed frame size. The Ethernet frame size may vary from 64
Bytes to 1518 B ( with 46 to 1500 bytes of a payload). In some technologies Ethernet
jumbo frames of 9600 Bytes are also supported.
Collision detection build into the Ethernet protocol allows on shared LAN networks
better efficiency than token-based technologies such as Token Bus or Token Ring. With
full duplex connections the speed of connection effectively doubles ( counting incoming
and outgoing traffic)
15 2013 Predictions: Mobile backhaul evolution in 2013 and beyond. Inhttp://www.rcrwireless.com/20130122/network-infrastructure/2013-predictions-mobile-backhaul-evolution-2013-beyond. Retrieved 12.15.2014.
Provisioning and Operations- Ethernet is connectionless technology. It simplifies
greatly the provisioning process, in comparison to network technologies such as ATM,
TDM, or SONET requiring establishing the connection between the end points of the
service. One may also overprovision the service ( which is impossible with connection
oriented technologies (TDM, SONET). Over-provisioning leads to the more efficient use
of the physical resources and decreases the cost per bit of the service. New developments
in the Ethernet standards introduced robust protection architectures (G.8031, G.8032
Architectures16) that match the Carrier protection levels of SONET/SDH networks ( in
particular sub-50ms convergence in linear and ring topologies).
With well defined services characteristics and interface specifications by MEF
standards, interconnections of different networks, connection of customers to the
METRO networks and specification of custom services are greatly simplified. New
SOAM standards (Y.1731 and IEEE 802.1ag17) implement in the Ethernet the OAM
features comparable to SONET or MPLS networks, making the Ethernet services viable
option for the Carrier Networks.
A Case in Point - Mobile Ethernet Back HaulingThe impact of Ethernet technologies on the services, demise of the ‘old’ networks
and the explosive growth of the Ethernet services is probably best witnessed on the
specific example. This example for us will be back-hauling of the wireless traffic, the
evolution that fueled unprecedented growth of Ethernet networks in METRO in late 2010
and beyond18.
The MEBH service is a service based on Ethernet technology standards designed
to carry or transport traffic between cell sites and Mobile Transport Switching Office
(MTSO) locations19. In the architecture of wireless networks, MEBH denotes the first
segment of the network connecting the cell site facilities with the MTSO equipment.
16 ITU-T G.8031/Y.1342 Ethernet Linear Protection Switching. 2011. ITU-T G.8032 : Ethernet ring protectionswitching.201217 ITU-T Y.1731 : OAM functions and mechanisms for Ethernet based networks. 2013. IEEE 802.1ag - ConnectivityFault Management. 2007.18 The section written based on Krzanowski R. Metro Ethernet Services for LTE Backhaul. ArtechHouse, Boston,2013.19 “Mobile or wireless backhaul is the portion of a wireless network that connects information traveling from a wirelesstower to a mobile switching center.” . Byme, D. What Is Mobile Backhaul. Accessed on the Web on March 23, 2012, athttp://www.ehow.com/facts_7406132_mobile-backhaul_.html.
Figure 1.9 presents a high-level view of wireless networks and the location of the MEBH
network segment in the Metro Ethernet Network. In most cases, MEBH is restricted to
the metro area, usually understood as an area covered within a radius of 120 miles or less.
The network providing the Ethernet service in the metro area is usually referred to as the
Metro Ethernet Network or MEN. Of course, such a definition is valid for urban areas. In
rural areas, MEBH may be extended over larger distances. In addition, the definition of
the service does not exclude multi-provider networks in the metro area; quite often, in
some locations the MEBH service is offered by two or more providers as a single
provider does not cover the entire service area.
Wireless backhaul may be differently understood by people depending on their
scope of responsibilities. Some may extend mobile backhaul to include the radio
equipment at the cell site as well as the equipment at the MTSO location. The definition
of the MEBH service is fairly loose in specifying the exact demarcation points. However,
the discussion in this book is limited to the segment of the wireline network between
interfaces facing the cell site on one end and the MTSO on the other end and remaining
under the management of the MEBH provider.
Figure 1.9 High-Level View of MEBH Service in MEN
The radius of the MEBH service is dictated by the wireless technology
requirements that set certain strict performance objectives for the traffic transport
between the cell site and the MTSO equipment. These performance objectives would be
impossible to support over large areas without a loss of the quality of the wireless service.
Several years ago, wireless carriers realized that their main service would evolve
from carrying voice to carrying data. Hand-held devices would become terminals for
voice, video, and text, supporting the full range of Internet services, evolving into
full-fledged multifunction data terminals demanding high bandwidth communication
channels. At the same time, the number of different applications that could be
implemented on the hand-held devices would grow beyond the wildest projections. This
evolution of cellular traffic would require explosive growth in wireless network capacity
as well as in all the segments of the transport networks carrying the wireless traffic20.
That was the reality to come.
In these early days, backhauling from the cell sites to the MTSO was
accomplished by using T121 or similar technologies. This architecture is presented in
Figure 1.10.
Figure 1.10. T1-Based Backhauling Design
T1 had its boom days, but with the limited capacity, cost, and maintenance
problems, this technology was not suited to support the future growth in capacity foreseen
by wireless providers. In the second half of the first decade of the 2000s, the diagram
presented in Figure 1.11, spelling doom to this method of carrying traffic, was circulated
on all conference presentations and copied in a multitude of industry magazines. The
diagram showed that with the T1 transport technology and the explosive growth in
20 It may come as a surprise to many that the biggest segment of the wireless services infrastructure is composed of thewireline networks.21 “The T1 is what telephone companies have traditionally used to transport digitized telephone conversations betweencentral offices. The bandwidth of a T1 is commonly known to be 1.544Mbps. This represents the maximum bit carryingability of a T1. The overhead necessary to frame a T1 is 8Kbps. Therefore, the total usable bandwidth is 1.536Mbps, orthe equivalent of 24 DS-0 channels. A single DS-0 has a bandwidth of 64Kbps and is designed to carry a digitizedtelephone call. Today, T1 technology is being used in private and publicnetworks to carry both voice and data traffic”. Pulse Technical Handbook Series - T1 Networking Made Easy. PulseInc., 2004. Accessed on the Web on May 19, 2012, at http://www.pulsewan.com/data101/pdfs/t1basics.pdf.
wireless traffic the cost of providing backhauling (BH) transport with T1s would outstrip
the revenues from the new data services.
A new method for transporting traffic from cell sites to MTSO locations needed to
be found, a method mature in the market, more efficient, more reliable, and cheaper.
Although with T1s wireless providers were able to carry several megabits of traffic from
a single cell site to an MTSO, marketing predictions professed the need for 20, 50, and
100 Mbps or more. And this was only the beginning, we were told. With this scenario
T1s were clearly out of the game. Of course, nobody in the sane mind believed in these
marketing curves predicting an exponential growth. But. Surprisingly, this prediction,
contrary to many other futuristic claims made in the past on a variety of topics, did
become reality22.
Figure 1.11. The “Doom” Scenario for Wireless Evolution (Circa 2007-2008)23
The choice of technology for MEBH was Ethernet or Carrier Ethernet (CE),
meaning the Ethernet network technology deployed for carrier-grade networking services.
Ethernet services have been maturing since the nineties (Ethernet is last-century
technology!) evolving from intra-office (enterprise) technology into Carrier Grade
networks. Ethernet services were cheaper, more reliable (than traditional T1s) if correctly
engineered and seemingly ubiquitous in metro areas. The problem was that although
network providers had extensive experience in using Ethernet as a carrier technology, for
mobile providers Ethernet was a new game. Yet the idea was born.
22 See for example Internet Collapses and Other InfoWorld Punditry. Robert M. Metcalfe . 198223 Carrier Ethernet for Mobile Backhauling. MEF. April, 2008.
The last few years (2010-2014) have witnessed explosive growth in Ethernet backhauling
services beyond even the most radical predictions. Looking toward the future, these
predictions still defy any previous estimates24. Some analysts predict that the combined
market of new wireless gadgets and applications ranging from all kinds of smartphones,
tables, and e-book readers to a plethora of online applications such as Internet radio,
video, TV, and many others probably yet unknown will increase the demand for backhaul
10x by 201625.
The actual evolution of the MEBH networking architecture shows the process of
adaptation of ‘legacy’ networks to the new BH paradigm - Ethernet - Figure 1.12
Figure 1.12 Evolution of Mobile Back-hauling architecture.
In the initial phase the mobile service providers maintained two network connections -
the connection over the legacy transport (A) and the connection over the Carrier Ethernet
(B). It was quite often the case that the MTSO and RAN BS equipment did not support
Ethernet UNIs. In this case the interfaces to the Carrier Ethernet networks was
accomplished using the GWR with legacy to Ethernet interfaces. The support for the
legacy network was also dictated by the fact that the mobile operators wanted to grow the
24 Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2010–2015. Accessed on the Web onNovember 10, 2011, athttp://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/ns827/white_paper_c11-520862.html25 Golstein, P. Study: U.S. Mobile Backhaul Demand to Grow Nearly 10x by 2016. Accessed on the Web on March 22,2012, at http://www.fiercewireless.com/story/study-us-mobile-backhaul-demand-grow-nearly-10x-2016/2012-03-13.
experience with the Ethernet services, they did lack initially. In the next phase the mobile
providers, after gaining some experience with the Ethernet services decommissioned the
legacy networks and used only the Carrier Ethernet service, in some cases still with
GWRs conversion services (C). Finally, the MTSO equipment and RAN BS were
upgraded to the full Ethernet UNI functionality, GWRs could be decommissioned and the
MTSO and RAN BS interfaced with Ethernet UNIs directly into the Carrier Ethernet
METRO network, achieving the end to end Ethernet service.
Evolution of EthernetWe may say that at the beginning there was the ALOHA network and Norman
Abramson.
ALOHA NetworkThe ALOHA network was the first practical implementation of the shared
communication channel ( with multiple access protocol)- the key element of the Ethernet
technology in its early days. The ALOHA network or ALOHA protocol was designed
in 1968 (with implementations in early 70ths) to support communication between the
central time-shared computer located at the main campus of the University of Hawaii and
distributed user terminals.The design was developed by Norman Abramson and his team
( Frank Kuo, N. Gaarder and N. Weldon)26. The design used two communication
channels one out-band and one in-band and the network was configured in the star
architecture. The central hub broadcasted messages to all the clients over the outbound
channel; the clients sent messages to the hub over the inbound channel.
The inbound channel was shared; the clients did not have to negotiate when to send
the message. Each message from the client was acknowledged by the hub with the special
message to the client. In case messages from different clients collided ( as the
communication channel was shared) the client did not get the acknowledgment, assumed
the collision, waited a certain amount of time (back-off) and attempted to resend the
26 R. Binder; N. Abramson; F. Kuo; A. Okinaka; D. Wax (1975). "Proc. 1975 National ComputerConference". AFIPS Press. Franklin F. Kuo (1981-08-11). "Computer Networks-The ALOHA System".Journal of Research of the National Bureau of Standards Vol.86, No.6, November-December 1981.Breyer Robert and Sean Ripley. Switching, Fast, and Gigabit Ethernet. 3rd ed. Macmillan TechnicalPublishing. USA. 1999.
message. Such acknowledgment/re-transmission scheme in which the clients did not have
to negotiate when to send the message greatly simplified the network and consequently
the client’s hardware. The second outbound channel was used to broadcast messages to
all clients; messages had the ‘client specific address’ allowing for the selective reception
of the messages from the hub.
Two variants of the ALOHA network were developed- Pure ALOHA and Slotted
ALOHA. In the Pure ALOHA design the clients did not check whether the channel was
busy and could send the message any time. The acknowledge/back-off/re-transmission
algorithm was used to provide the reliable communication. This configuration resulted in
random collisions limiting the capacity of the channel. The slotted ALOHA introduced
discrete time slots ( it required the global time synchronization which the Pure ALOHA
did not); client could only sent the data at the beginning of each time slot. This method
reduced the collisions and increased greatly the throughput of the channel ( from about
10% for Pure ALOHA to the theoretical 36.8 % for slotted ALOHA)27. However, the
multiple access ALOHA protocol could never achieve 100% throughput of the
transmission channel.
Figre 1.13 Performance of the Pure Aloha and Slotted ALOHA in the function of the
traffic load G28.
For the Pure ALOHA assuming G as a offered mean load (frames per time frame) the
throughput S is expressed as:
S= Ge -2G
27 Tanenbaum, A.S. Computer Networks, 4th Ed. Prentice Hall, Upper Saddle River. 2003.28 http://en.wikipedia.org/wiki/ALOHAnet. Retrieved 12.12.2014.
The maximum throughput (theoretical) occurs at G=0.5 ( S=1/2e) equal to about
(practically) 0.184.
For the Slotted ALOHA the throughput S is expressed as:
S= Ge-G
The maximum throughput (theoretical) occurs at G=1 ( with S=1/e) equal to about
( practically) 0.36829.
The speed of the ALOHA network was eventually 9600 bps ( upgraded from 4800
bps). Packets consisted of a 32-bit header, 16-bit priority check field, and up to 80 bytes
of data and a 16-bit check word for the data part of the packet. The header contained the
unique client identification information, so the client accepted only the data sent to him in
the broadcast message.
The key concepts in the ALOHA network were the use of two channels (outbound
and inbound) and the random access of the clients for the inbound transmission. This
ideas was to be reused in the concept of Ethernet30.
Xerox PARC YearsIn 1972 Bob Metcalf was working in the Computer science Laboratory of the Xerox
Palo Alto Research Center on the network to connect the Xerox computers to the
Arpanet31. He read Abramson’s paper from 1970 on the ALOHA network and realized
that the throughput of the network may be improved to reach 100% capacity. In May of
1973 the first implementation of such a network was demonstrated. It run at 2.94 Mbps32
and used its own multiple access protocol algorithm known as CSMA/CD ( Carrier
Sense Multiple Access with Collision Detection). In 1976 100 nodes were connected over
the coaxial cable at the Xerox research Center PARC. This year the famous paper
29 Tanenbaum, A.S. Computer Networks, 4th Ed. Prentice Hall, Upper Saddle River. 2003.30 The interesting twist in the Ethernet story is that the ALOHA network by the size of its service area was not theLAN but WAN network.31 The Advanced Research Projects Agency Network (ARPANET) was one of the world's first operational packetswitching networks, the first network to implement TCP/IP. It begun operations 1969. See also Roberts, Larry(November 1978). "The Evolution of Packet Switching". Proceedings of the IEEE 66 (11): 1307. And · Roberts,Larry (September 1986). "The ARPANET & Computer Networks". ACM.32 This speed was the result of the clock in the PARC computer that clocked the interfaces with a pulse every 340nanoseconds.
describing the Ethernet architecture was published33. In 1977 Robert M. Metcalfe, David
R. Boggs, Charles P. Thacker, and Butler W. Lampson received U.S. patent number
4,063,220 on Ethernet for a "Multipoint Data Communication System With Collision
Detection."
At this point in time Ethernet was defined by the following features; the capacity of
2.94 Mbps, the CSMA/CD algorithm, it was running over the 1000-meter thick coaxial
cable (or ThickNet), similar to RG-8/U coax cable connecting (in 1976) 100 nodes. The
nodes NICs were connected to the external transceiver unit with an AUI34 connector that
was in turn connected to the coax cable. The Ethernet Frame included 8 bits field for
Destination Address ( one byte) , 8 bits field for source Address ( one byte) around
4000 bits filed for the data, and 16 bits for the Checksum (two bytes)35. The network
design for the first Ethernet network was called the shared bus topology. The Thick
Ethernet Cable is presented on Figure 1.14.
Figure 1.14 Thick Ethernet cable36 (RG-8 IEEE 802.3 10Base5).
33 Robert M. Metcalfe and David R. Boggs (July 1976). "Ethernet: Distributed Packet Switching for Local ComputerNetworks". Comm. of the ACM 19 (7).34 Attachment Unit Interface35 Robert M. Metcalfe and David R. Boggs (July 1976). "Ethernet: Distributed Packet Switching for Local ComputerNetworks". Comm. of the ACM 19 (7).36 Thick 50 ohm Ethernet trunk cable http://www.iec.net/cab-trunk.html
The 1976 Ethernet network with its components is shown in Figure 1.14. This
implementation of the Ethernet is sometimes referred to as “Experimental Ethernet”.
Figure 1.15 Ethernet Architecture from 1976 Metcalf and Boggs’ paper.37
After XeroxThe subsequent story of the Ethernet is the story of rapidly evolving standards,
standards wars, dead technologies, and enthusiasm in the market that made Ethernet
ubiquitous and the preferred choice over all other competing solutions.
In the Fall of 1979 DEC, Intel and XEROX (The first letters of these corporations
created the acronym DIX) begun to work on the Ethernet Specification. It was published
the next year as the Ethernet Blue Book or DIX; this was DIX V 1.0 standard that offered
the speed of 10 Mbps with the Ethernet’s thick trunk cable scheme38. In 1982, after two
years of work the DIX standard was refined as DIX V 2.0 and published in 1982 as The
Ethernet Version 2.039 or Ethernet II as it is usually known. The specifications differed
from the “original Ethernet from 1976. The maximum length was extended to 2500
meters with the segments of up to 500 meters. The preamble increased from 1 bit to 64
bits, the address filed had 48 bits each, and the CRC filed had 32 bits. Both Ethernet
37 Metcalf, Boggs (1976).38 Digital Equipment Corporation, Intel Corporation and Xerox Corporation (30 September 1980). "The Ethernet, ALocal Area Network. Data Link Layer and Physical Layer Specifications, Version 1.0". Xerox Corporation. Retrieved2014-12-12.39 Digital Equipment Corporation, Intel Corporation and Xerox Corporation (November 1982). "The Ethernet, A LocalArea Network. Data Link Layer and Physical Layer Specifications, Version 2.0". Xerox Corporation. Retrieved2014-12-12.
(experimental and DIX) specifications used Manchester encoding.
The following table summarizes the features of both Ethernet specifications40..
Table 1.5 Experimental and DIX Ethernet Specifications.
In 1981 the IEEE decided to standardize the LAN technologies and created the
subcommittee to develop an international standard based on DIX specifications. The
IEEE standard was created within the IEEE Local and Metropolitan Networks
(LAN/MAN) Standards Committee, which identifies all the standards it develops with the
number 802. The subcommittee was designated as 802.3. The specifications were
completed in 1983 as a draft and became the IEEE 10Base541 standard in 1985. The
IEEE 802.3 is not exactly the same as DIX v 2.0 but the differences are minor. The IEEE
802.3 specifications allow backward compatibility with the systems build according to
earlier DIX specifications. We may assume safely that all of the Ethernet equipment built
after 1985 is based on the IEEE 802.3 standard. Subsequent supplements to the IEEE
802.3 specifications have been identified by a small letter or letters added to the number.
40 John F. Shoch; Yogen K. Dalal; David D. Redell; Ronald C. Crane (August 1982). "Evolution of the Ethernet LocalComputer Network". IEEE Computer 15 (8).41 How to read an Institute of Electrical and Electronics Engineers (IEEE) shorthand identifier. The "10" in the mediatype designation refers to the transmission speed of 10 Mbps. The "BASE" refers to baseband signalling, which meansthat only Ethernet signals are carried on the medium. The "T" represents twisted-pair; the "F" represents fiber opticcable; and the "2", "5", and "36" refer to the coaxial cable segment length (the 185 meter length has been rounded up to"2" for 200). http://searchnetworking.techtarget.com/definition/10BASE-T.
ExperimentalEthernet
DIX V 1.0
Date Rate 2.94 Mbps 10 Mbps
Maximum End-to-end rate 1000 m 2500 m
Maximum Segment Length 1000 m 500 m
Encoding Manchester Manchester
Coax Cable Impedance 75 ohms 50 ohms
Coax Cable Signal Levels 0 to +3 V 0 to -2 V
Transceiver Cable connectors 25- and 15- pins Dseries
15-pin D series
Length of preamble 1 bit 64 bits
Length of CRC 16 bits 32 bits
Length of address fields 8 bits 48 bits
Major revisions of the IEEE 802.3 have been identified with the revision year.
Xerox did nothing to further the Ethernet technology and in 1979 Bob Metcalf left
the company. In 1979, with Howard Charney, Bruce Borden, and Greg Shaw he jointly
founded 3Com with the purpose of productizing the Ethernet technology. 3Com
specialized in building Ethernet adaptor card for early computer systems such as Apple,
LSI-11, IBM PC, and VAX-11. In 1982 3Comm developed the NIC card - the
EtherLink for the IBM PC introducing the Thin Ethernet which was much cheaper, easier
to install and did not need the bulky, external transceiver unit. Figure 1.16 shows the thin
Ethernet cable.
Figure 1.16. Thin Ethernet cable42 (RG58 Thinnet 10Base2 Coax Cable- IEEE 802.3
10Base2).
The Thin Ethernet (cheapernet, thinwire, thinnet) was named 10Base2; for 10 Mbps
throughput and 200 m segment size ( in fact 185 meters). The Ethernet market for PCs
was growing beyond any predictions and in 1983 3Com filed its first public offering of
stock. The Thin Ethernet was defined as a part the IEEE 802.3a standard in 198443.
Both 10Base5 and 10Base2 networks supported bus topology. The next milestone in
the Ethernet evolution was the development of the start topology and introduction of the
UTP wire. The start topology ( tree or hierarchical topology) was easier to install,
maintain, configure and troubleshoot, lowering the operational cost of the whole network.
42 http://www.showmecables.com/product/belden-9907-rg58-thinnet-10base2-ethernet-coaxial-cable-per-ft.aspx43 In 2012 3Com was acquire by Hewlett-Packard and ceased to exist.
It was initially developed by Intel ( with the partnership from AT&T and NCR) in 1983
as StarLAN44. StarLAN was defined by the IEEE 802.3e standard in 1986 as the 1Base5
version of Ethernet and in 1987 StarLAN was patented by AT&T as US Patent
4674085. The advantage of StarLan was that it run over the telephone, twisted pair wire,
so it could reuse existing telephone installations. However, it was slow delivering only
1Mbps. This cemented demise of the bus network and StarLAN itself.
In 1987 SynOptics Communications introduced LattisNet delivering 10 Mbps
throughput over an unshielded twisted pair in a start topology know as Ethernet 10BaseT.
Two years earlier in 1985 the Xerox engineer (Eric Rawson) demonstrated that Ethernet
can be run over the fiber, but only in a start topology. The first LattisNet hub for fiber
optics and shielded twisted pair (STP) was shipped in 1987. LattisNet was called the
Ethernet-over-telephone-wire technology and was hailed as price and technology
break-through as it proved the superiority of the UTP over the thinnet coaxial cabling.
For the record, the tests performed by Novell showed the LattisNet thought at
173.5Kbyte/sec and thin Ethernet for the same conditions to be 168.2Kbyte/sec45. In
1990 the IEEE approved the 802.3i/10Base-T version of Ethernet over the UTP cable.
While StartLAN never took off as the LAN technology, it has to be credited with the
introduction of the novel architecture (start topology) replacing the bus design of earlier
Ethernet.
The next milestone in the Ethernet evolution was the introduction of the Ethernet
Switch in early 1990s. The invention of the Ethernet switching is credited to Kaplana46.
Kaplana in 1989 introduced into the market a multi-port Ethernet switch with 7 ports The
Kaplana EtherSwitch EPS-700. The Kaplana switch used new cut-through rather than
store-and-forward technology for packet processing, in-hardware processing logic,
allowed for multiple simultaneous data transmission flows, increasing the performance
of the switch by the order of magnitude. The Kaplana switch was not positioned as a
bridge but as a device boosting the LAN performance. It created a new category of the
44 Mary Petrosky (June 9, 1986). "Starlan nets: Chip set chips away at Interface cost". Network World 3 (14). p. 4.45 Eric Killorin (November 2, 1987). "LattisNet makes the grade in Novell benchmark tests" 4 (44). Network World.p. 19.46 Kaplana was co-founded by Vinod Bhardwaj and Larry Blair.
networking devices - the Ethernet network switch and it soon became the standard
equipment in the LAN space. The basic parameters of the first Kaplana switch are
presented in Table 1.6. It is worth having this table in mind when reviewing the
specifications of the to-date networking equipment to see how far did the industry
advanced over last 20 years. In 1994 Kaplana was acquired by Cisco.
Ports 7ForwardingMethods
Cut-through packetforwarding
Throughput 30 MbpsLatency 40 μsBuffer 256 packets
Table 1.6 Kaplana EtherSwith-700 Specifications
In 1993 Kaplana developed another break-through technology; the full duplex
Ethernet. Full-duplex allows to simultaneously transmit and receive increasing the
bandwidth two-fold. Of course, this is only possible in the start topology with
point-to-point links. In the bus topology Ethernet operates in the half-duplex mode. In the
full-duplex mode there is no possibly of packet collision. As there is no need for the
CSMA/CD detection the limitations of the distance due to the collision detection are no
longer existing and the link distance depends only on the strength of the signal in the
transport medium used. In 1997 the IEEE published 802.3x full duplex/ flow control
Ethernet specifications.
The next step in the Ethernet ascend was the development of 100Mbps links called
Fast Ethernet. Fast Ethernet in fact refers to the several standards that carry traffic with
the rate of 100 Mbps. Fast Ethernet can use the UTP wire or an optical cable in the start
topology. Fast Ethernet is referred to as 100Base-X where X stands for FX, SX, BX, T4,
LX10, or TX depending on the physical media used. The IEEE published the 100Base-T
standard in 1995 as 802.3u.
This is how 100 Mbps Ethernet was born. In 1992 the need for the faster than 10
Mbps Ethernet were discussed in the IEEE 802.3 committee. Lack of agreement among
the attending parties caused the creation of the splinter group The Fast Ethernet Alliance47.
The group’s objective was to develop on the basis of existing Ethernet standards the
specifications for 100Mbps Ethernet (the competing solution, based on a completely new
MAC method, came from Hewlett-Packard). In 1993 the Fast Ethernet Alliance
publishes its 100Base-T document and Grand Junction Networks begins to ship its Fast
Ethernet hubs and NICs. Grand Junction Networks were founded in 1992 with explicit
goal to develop and market high-speed Ethernet equipment48. By 1994 more vendors
jointed the Fast Ethernet Alliance shipping new products and the market for Fast Ethernet
was exploding.
Subsequent technology milestones that define the development of Ethernet are 1000
Mbps Ethernet published in 1998 as the IEEE 802.3z (1000Base-X 1Gbps standard),
development of 10Gbps Ethernet(IEEE 802.3ae published in 2002 with several
subsequent revisions), development of 40Gbps and 100 Gps Ethernet (802.3ba published
in 2010) specifications and, in the future 400 Gbps Ethernet (802.3bs expected to be
completed in 2017).
Table 1.7 provides the comparison of basic characteristics of Ethernet 10Base5,
10Base2, 10Base-T, 1Base5, 100Base -X and 1000Base-X.
Designation Supported Media Maximum SegmentLength
TransferSpeed Topology
10Base-5 Coaxial 500m 10Mbps Bus10Base-2 ThinCoaxial (RG-58 A/U) 185m 10Mbps Bus
10Base-T Category3 or above unshieldedtwisted-pair (UTP) 100m 10Mbps
Star,using either simplerepeater hubs orEthernet switches
1Base-5 Category3 UTP, or above 100m 1Mbps Star,using simplerepeater hubs
100Base-TX Category5 UTP 100m 100MbpsStar,using either simplerepeater hubs orEthernet switches
47 The Fast Ethernet Alliance was funded in 1993 by the consortium or Ethernet vendors including Grand Junction,Intel, LAN Media, SynOptics, Cabletron, National Semiconductor, Sun Microsystems, and 3Comm. Hewlett_Packardand AT&T did not join.48 Grand Junction Networks. http://www.fastcompany.com/33233/grand-junction-networks.
100Base-FX Fiber-optic- two strands of multimode62.5/125 fiber
412 meters(Half-Duplex)
2000 m (full-duplex)
100 Mbps
(200 Mb/sfull-duplexmode)
Star(often onlypoint-to-point)
1000Base-SX Fiber-optic- two strands of multimode62.5/125 fiber 260m 1Gbps
Star,using buffereddistributor hub (orpoint-to-point)
1000Base-LX Fiber-optic- two strands of multimode62.5/125 fiber or monomode fiber
440m (multimode)5000 m(singlemode)
1GbpsStar,using buffereddistributor hub (orpoint-to-point)
1000Base-T Category5 100m 1Gbps Star
Table 1.7 Comparison of characteristics f Ethernet from 10Base5 to 1000Base-X49.
In the wide adoption of Ethernet the IEEE standard called Ethernet in the first mile or
EFM was of critical importance. The standard IEEE 802.3ah-2004, later to be included in
the IEEE 802.3-2008 defined the Ethernet technology to be used for access between the
customer’s premises and the telecommunication company access facilities. The EFM
facilitated the deployment of the pure Ethernet transport rather than Ethernet over ATM
or FR.
In addition to speed increases the development in the service and control planes
significantly contributed to the Ethernet adoption. These include the development of
VLAN construct with its ability to signal QoS and create virtual connections (EVCs)50,
development of PWEs for Ethernet transport51, the development of protected ring and
linear architectures52, and the development of the whole suit of SOAM facilities53. The
introduction of EVCs allowed for customer specific services, high-granular QoS, and
segmentation of the control plane; making Ethernet services more sophisticated and
managed (2003). The specification of PWEs for the Ethernet transport (2006) allowed to
49 10BaseT 10BaseF 10Base2 5-4-3 rule 10Base5 100BaseFX 100BaseT4 100BaseTX.http://computernetworkingnotes.com/network-technologies/10base-ethernet.html50 IEEE 802.1Q is the most widely used implementation of the VLAN Ethernet. There is also the Cisco proprietary ISLprotocol ( Inter-Switch Link). IEEE 802.1Q was published in 1998 and had several revisions in 2003 (802.1Q-2003),2005 (802.1-2005), and 2011 (802.1Q-2011).51 IETF developed a series of standards in the area of pseudo-wire and MPLS. Probably the most important forEthernet are RFC 3985 (2005) Pseudo Wire Emulation Edge-to-Edge (PWE3) Architecture and RFC 4448 (2006)Encapsulation Methods for Transport of Ethernet over MPLS Networks.52 ITU-T G.8032 Ethernet Ring Protection switching architecture developed between 2010 and 2013. ITU-T G.8031Ethernet Linear Protection switching developed between 2006 and 2013.53 ITU-T Y.1731 – OAM Functions and Mechanisms for Ethernet based Networks developed between 2006 and 2013and IEEE Std 802.1ag-2007 Virtual Bridged Local Area Networks, Amendment 5: Connectivity Fault Management
extend the Ethernet over MAN and WAN in the same time providing robust MPLS
protection and resiliency mechanisms, scaling and QoS. The Ethernet PWEs provided the
entry of the Ethernet service into the Carrier world. Protected linear (2006) and ring
(2010) architectures introduced SONET-like protection design for Ethernet into the MAN
and fostered the development of so called connection -oriented Ethernet or COE service;
the technologies that opened MAN and WAN networks for Ethernet LAN. And the
development of complex SOAM facilities (2006-7) offering the SONET/SDH -like
complex control and management plane for Ethernet services contributed to the
establishment of Ethernet as Carrier Grade network technology. The specifications for
these technologies were developed by IETF, ITU-T and IEEE standard organizations.
The creation of Metro Ethernet Forum, in 2001, to foster the convergence of Ethernet
services, had difficult to overestimate impact on the Ethernet world. Metro Ethernet
Forum (MEF) was founded with the specific task of defining Ethernet Services. While
not technical in the understanding of IEEE work, the work of MEF was critical to
wold-wide adoption of Ethernet as a service medium by providing the unified
terminology and definition of service types, interfaces, NID equipment, performance
measures, OAM specifications, equipment and service certification, and architecture. The
MEF terminology and conceptualization of service features became the lingua franca of
the Ethernet service providers and Ethernet equipment vendors. In 2011 MEF launched
CE 2.0 Certification Program introducing the carrier features into the Ethernet service.
The milestones in the Ethernet technology development in IEEE standards are
summarized in Table 1.5. The MEF standards milestones are presented in Figure 1.17.
Ethernet
StandardDate Description
Experimental
Ethernet1973 2.94 Mbit/s (367 kB/s) over coaxial cable (coax) bus
Ethernet II
(DIX v2.0)1982 10 Mbit/s (1.25 MB/s) over thick coax.
IEEE 802.3
standard1983
10BASE5 10 Mbit/s (1.25 MB/s) over thick coax. Same as
Ethernet II (above) except Type field is replaced by Length,
and an 802.2 LLC header follows the 802.3 header. Based on
the CSMA/CD Process.
802.3u 1995100BASE-TX, 100BASE-T4, 100BASE-FX Fast Ethernet at
100 Mbit/s (12.5 MB/s) w/autonegotiation
802.3x 1997 Full Duplex and flow control
802.3z 19981000BASE-X Gbit/s Ethernet over Fiber-Optic at 1 Gbit/s
(125 MB/s)
802.3ae 2002
10 Gigabit Ethernet over fiber; 10GBASE-SR, 10GBASE-LR,
10GBASE-ER, 10GBASE-SW, 10GBASE-LW,
10GBASE-EW
802.3ah 2004 Ethernet in the First Mile
802.3bm 2014 100G/40G Ethernet for optical fiber
802.3bs ~ 2017400 Gb/s Ethernet over optical fiber using multiple 25G/50G
lanes
Table 1.5 Evolution of Ethernet technology in IEEE 802.3 working group standards54.
54 Adapted with some changes from http://en.wikipedia.org/wiki/IEEE_802.3. Retrieved 12.12.2014
Figure 1.17 Development of Ethernet Service Specifications in MEF55.
55 An Overview of The Technical Work of MEF. MEF Reference Presentation 2011.http://metroethernetforum.org/carrier-ethernet/presentations Retrieved 12.14.2014.