Metro Ethernet ripples out to the WAN

This is turning out to be the year of metro Ethernet. The idea of a ubiquitous Ethernet global network has been in the cards since ATM lost out to Gigabit Ethernet (GbE) in the LAN techno-war. Now, even the more sane analysts such as Michael Howard, co-founder of Infonetics Research (San Jose CA), are getting excited about its chances of reaching the wider area. "In the next 10 years, Ethernet will inexorably take over the metro," says Howard. "Of course, there will never be a wholesale change because of the SONET/SDH installed base, but every year Ethernet will be accounting for a larger portion of metro capital expenditures. Between 2002 and 2006, Ethernet will make major inroads into metro telecom equipment spending, accumulating $18.9 billion."

Timely then that Broadband-Testing has run its eye over the recent Metro Ethernet Forum (MEF) interoperability demonstration at last June's SuperComm. The full report will be available shortly at www.broadband-testing.co.uk. The SuperComm interop demo was the largest and most comprehensive service interoperability demonstration of metro Ethernet services to date. In total, 28 MEF members took part in creating an end-to-end metro Ethernet networ, to support the E-Line and E-LAN services.

As defined by the MEF, E-Line provides a point-to-point Ethernet virtual connection (EVC) between two subscribers. E-LAN provides multipoint connectivity between two or more subscribers (known as UNIs). From a subscriber standpoint, an E-LAN service makes the Metro Ethernet network look like a LAN.

The demo was built around a 10-GbE core based on Cisco Systems' and Extreme Networks' switch technology. Supporting that were a number of Ethernet aggregation and access devices connecting a variety of hosts running applications that varied from Webcam-enabled instant messaging to IP telephony and video streaming over Ethernet point-to-point and point-to-multipoint services. Primary goals of the test were to:

  • Prove that metro Ethernet services can successfully extend packet delivery from the LAN to service-provider network access, aggregation, and core layers.
  • Show that metro E-Line and E-LAN services are carrier grade solutions that provide seamless end-to-end connectivity to hosts.
  • Demonstrate that metro Ethernet products from multiple vendors can fully interoperate at all layers of the service-provider network.

Beyond the SuperComm demo and the MEF, it's clear that most of the networking vendors are looking to get into the metro Ethernet market as soon as possible, which is a sign in itself of the way the industry believes Ethernet is moving into the areas previously reserved for other technologies. LAN-oriented switch vendors such as Foundry Networks (shortly to become an MEF member) are investing heavily in metro Ethernet-oriented products. Similarly, Hitachi—heavily involved in storage and mainframe technology—has entered the metro Ethernet market. This broad range of market entries, again from an interoperability viewpoint, is a healthy sign for the future of metro Ethernet technology, since it is not being dominated by any specific vendor or market sector.

Closely related to the MEF is the Ethernet in the First Mile Alliance, whose goal is to establish Ethernet (the IEEE p802.3ah taskforce) as the technology of choice for the first/last mile of network to feed residential or office users directly, working hand-in-hand with metro Ethernet feeds. The technology is designed to run over copper or fibre cable, meaning in this case relatively low-speed Ethernet copper-wire-based DSL-type deployments are supported, as are higher-speed alternatives over copper and fibre. Again, the key element is simplicity and ease of management, both aimed at drastically reducing the cost of deploying metropolitan broadband access. Companies such as Metrobility are looking to enable service providers to deliver what in theory are complex high-bandwidth metropolitan services with as close to a plug and play delivery mechanism as possible. This approach is leagues away from the ATM-oriented approach of DSL, for instance, that we have seen to date, where configuration and deployment are relatively complex, expensive, and time-consuming.

While time scales are still uncertain, what does seem absolute is Ethernet will become truly ubiquitous over the next few years—it is already in countries like South Korea, where government funding has allowed for mass delivery of broadband Internet access into homes, with Ethernet a key enabling technology. That should be only good news for users—be they business or residential—since it simplifies their world of IT and should mean more bandwidth for less money.

Recently, Broadband-Testing had a pure Ethernet-based DSL solution (DSLAMs and CPE) from Net To Net Technologies in our laboratories and were impressed with its ease and speed in setting up and overall performance. While historically all "mainstream" xDSL DSLAM solutions have been ATM-based, Net To Net opted for a pure Ethernet approach from day one. The ATM camps argue that Ethernet lacks the required scalability. Net To Net points to the same old arguments originally used against GbE, and look where that technology is now.

More important, Net To Net argues that the simplicity of the Ethernet solution brings deployment and management costs down significantly. The key point here is speed of return on investment for the service provider or enterprise deployment. In Europe, service providers have often blamed the cost of provisioning the service and the inherent management costs thereafter for the lack of resultant revenue. The basic cost of entry for an Ethernet-based solution is also significantly lower, though obviously that levels out more as port density increases through economies of scale.

Broadband-Testing also put the features to the test with a video over DSL trial; it ran extremely well. With ADSL now running up to 10.5-Mbit/sec downlink speed and Net To Net offering a 9.2-Mbit/sec SDSL (symmetrical bandwidth) solution, the Ethernet-based DSL route is certainly worth considering and will naturally work alongside metro Ethernet deployments as they become more widespread because of simple Ethernet compatibility.

Steve Broadhead is laboratory manager of France-based Broadband-Testing, a new independent test house offering dedicated testing, consultancy, and development facilities. He can be reached via the company's Website, www.broadband-testing.co.uk.

More in Packet Transport