Media Coverage

Riding the Wave

By Sean Buckley

Fast forward to 2007, and Ethernet is ready to ride a new wave as a transport mechanism.CET (Carrier Ethernet transport) combines the cost efficiency of Ethernet, the reliability of WDM and traditional SONET/SDH to deliver flexibility, efficiency and cost savings to network interconnections.

Then CET combines agile optics with Ethernet tunnel technologies such as PBB-TE (provider backbone bridging: traffic engineering) or T-MPLS (transport MPLS).

What’s driving this trend? Service providers are looking to leverage Ethernet’s cost efficiency with the resiliency of traditional SONET/SDH networks for wireless backhaul, DSLAM backhaul, and wholesale bandwidth applications. Still, there’s a lot of work to be done.

"Ethernet’s come a long way, but SONET/SDH networks provide a well-understood suite of network monitoring and control functions that Carrier Ethernet is just starting to deliver on," said Jason Marcheck, principal optical analyst for Current Analysis.

"Add to this the increasing demands that sophisticated services like VLANs, IP video, etc., place on OAM, and it becomes a complex solution. It means that CET solutions have to deliver the cost-effective scalability of Ethernet with the network control and monitoring features of SONET/SDH.

Things like PBT and end-to-end IP/MPLS are what vendors are starting to [rely on] to provide these network control features."

PBT/PBB-TE Captures Hearts, Headlines

One part of the new CET movement gaining the most headlines is the emerging PBT/PBB-TE standard effort. PBT (provider backbone transport), which will be standardized in the IEEE as PBB-TE, provides various enhancements to make Ethernet more deterministic in nature. PBB-TE can bring control to data paths within a large carrier network, enabling QoS and the ability to set aside specific paths for specific traffic types.

Providing a connection-oriented transport mechanism, PBT utilizes the existing IEEE’s 802.1ah standard PBB format for scaling and layering. What’s more, PBB-TE brings the connection-oriented characteristics and deterministic behavior of SONET/SDH to Ethernet. PBB-TE achieves this by disabling the concept of flooding/broadcasting and spanning tree protocol, enabling it to act like a traditional carrier transport technology.

"The basic advantage of PBT/PBB-TE is cost," said Michael Howard, principal analyst for service provider optical, routing and metro Ethernet at Infonetics Research. "If you can keep all of the transport at Layer 2, you have to buy Ethernet switches; you don’t have to buy routers. PBT allows you to build a total Layer-2 transport network with no routing. That and the operational simplicity are the benefits of PBT vs. using MPLS."

On the vendor side, the PBB-TE drive is coming from two angles: traditional optical vendors and Ethernet switching vendors. In addition to Nortel–one of the most vocal proponents of PBB-TE–Avici, Extreme Networks, Meriton Networks, Siemens and World Wide Packets have marshaled support for PBB-TE.

With much of the initial push around new CET formats coming from the vendor community, it’s clear there’s yet no winning approach.

Complementing the PBT/PBB-TE movement is Avici’s Soapstone solution. Poised as a virtual control plane for PBT, the Soapstone Provider Backbone Transport Controller can help smooth the service provider transition to PBT when ready.

Juniper, Cisco and Atrica, however, have taken a more neutral approach to the CET debate. Atrica, which has been offering connection-oriented Ethernet switches since 2000, feels as though the T-MPLS and PBT movements are quite complementary.

"The five attributes of the MEF vision (i.e., scalability, protection, SLAs, TDM integration and service management) are complemented by PBT and T-MPLS," said Unmesh Kukreja, director of marketing for Atrica. "The bottom line is both PBT and T-MPLS, regardless of which one wins in the market, have started to go into this connection-oriented paradigm."

MPLS Claims Early Lead

For all that PBB-TE promises, support for T-MPLS is just as strong. T-MPLS is designed to simplify and enhance MPLS by enabling transport-focused packet solution.

T-MPLS, based on the existing ITU-T G.805 layered transport architecture, supports Ethernet and legacy over client pseudowires. T-MPLS allows for end-to-end protected bi-directional LSPs in three ways:

• By removing elements that disrupt OAM integrity (e.g., PHP, ECMP);

• By adding OAM and switching functionality; and

• By configuration and monitoring via network management.

Proponents of T-MPLS, including Alcatel-Lucent, Ericsson and Fujitsu, argue that the T-MPLS approach, which provides simplified OAM&P processes, is farther along in the standards process, while PBT is just getting started.

Still, many in the PBB-TE camp remain undaunted.

Mick Reeve, an independent consultant and former chief architect on BT’s 21CN, says the PBB-TE group can leverage similar elements adopted by T-MPLS.

"In practice there’s quite a bit of work to do, and transport-MPLS is ahead, but–and it’s quite a big but–when you look what we have to do in both PBT and T-MPLS, there are some very minor tweaks to how the packets are routed," Reeve said. "The IEEE guys don’t expect that to be a big piece of work. What is a big piece of work is working out operations and maintenance issues."

The Optical Perspective

But Ethernet switching is only one part of the equation. Optical vendors, including Alcatel, ADVA, Nortel, Siemens and newer vendors Foundry and Meriton, are also incorporating CET functionality into their platforms.

Coming from the router perspective, Foundry Networks has integrated PoS (packet over SONET) interfaces on both its NetIron XMR Internet routers and NetIron MLX Series of metro routers.

Meriton Networks, for one, has made a big move on PBB-TE via a CET product upgrade roadmap for its existing 7200 OSP platform.

The vendor’s two-phase strategy will enable carriers to support GigE networks via sub-wavelength switching and grooming of nine GigE streams and integrating Ethernet gateway tunnels (i.e., PBB-TE and T-MPLS) with intelligent WDM.

"Carrier Ethernet transport is a very clean interface between the service layer and the transport layer," said Bill Gartner, Meriton COO.

"As an Ethernet handoff between those two layers, the service layer can do its job, which is to manage the services and not have to have specific knowledge about the transport layer. What it needs to know is that it can get a service between point A and Z and the transport layer can provide that capability."

Tom Nolle, president of CIMI Corp., sees value in this approach.

"The whole paradigm of networking changes if you commoditize the underlying property of the bit, and that’s what we’re doing with optics," he said. "What Meriton and others are saying is, if you can make Ethernet and lambdas almost philosophically interchangeable, then you can do all of your traffic engineering and all your high volume networking below IP where it can’t be attacked, where it’s not subject to [variability], and where its cost is lower."

Service Providers Take Notice

Even though it’s still early in the game, service providers are paying attention.

PBB-TE got a big boost last January when BT announced it would deploy PBT-enabled Siemens and Nortel Ethernet equipment in its 21CN network implementation.

BT, which will use Nortel and Siemens Ethernet platforms to transport its residential and business service traffic, sees that the implementation of PBT will complement its 21CN MPLS strategy (see sidebar: BT’s Ethernet Vision). Outside of BT, industry sources say Bell Canada and Verizon are also in favor of PBB-TE.

Tom Bechley, director of enterprise network engineering for Verizon, while not saying it is committed to deploying PBT, is investigating its prospects.

"From our perspective, we have already gone down a path and the problems proponents of PBT say they can solve are things we recognized all along and addressed them within the scope of current MPLS," he said. "You have to imagine vendors talk to us about this stuff, and we have been doing an exercise with PBT to see if there’s any forward-looking value."

If anything is certain in the emerging CET market (in whatever form it takes), service providers will be watching intently how Ethernet can provide them with a cost effective and flexible transport mechanism that has the same reliability as legacy circuit-based transport technologies.