BONDING THAT BINDS

Recent market metrics and the ongoing machinations of the Metro Ethernet Forum — whose carrier Ethernet certification program began earlier this year — leave little doubt about the increasing effectiveness with which telcos are breaking down the walls that traditionally have kept them from making an impact in the Ethernet market.

For many years, telcos watched the Ethernet market grow — and with it a lot of potential to do business with large corporate enterprises. In part, telcos were left out because they favored other technologies, such as ATM, to support enterprise services. But, there also was not a standardized way of delivering Ethernet over their existing TDM infrastructures that used plesiochronous digital hierarchy (PDH) signaling to transmit over DS-3, T-1, T-3 and E-1 circuits.

New technology standards are helping to resolve these lingering issues. Among the most well-known of those standards is the IEEE's G.HDSL Ethernet in the first mile (EFM) standard for delivering Gigabit Ethernet over copper. Passed in early 2004, the EFM standard addressed the fact that the vast majority of telcos' access loops are still copper based.

Later, in September 2004, more help came from the International Telecommunications Union, which addressed the issue of legacy signaling with its G.7043/Y.1343 virtual concatenation of PDH signals recommendation. The long-winded name belies a simple idea — and a familiar one to anyone familiar with virtual concatenation and generic framing in Sonet/SDH networks. In short, under the ITU circuit-bonding recommendation, multiple PDH circuits are bonded into a single virtual pipe better suited to support Ethernet or other high-bandwidth services.

During a critical time when telcos are trying to offer such services but are taking a very gradual approach to their IP migrations, this method of circuit bonding can give them another leg up in the competitive race to provide Ethernet services.

“It's another step along the path of bring Ethernet to customer sites that don't have access to fiber,” said Sterling Perrin, senior research analyst for optical networks at consulting firm IDC. “Ethernet over copper itself is a big application, and the ability to do a highly available Ethernet service over copper is that much more important. Ethernet is still relatively expensive, and it makes the Ethernet decision that much easier for a large corporate enterprise or even a small enterprise with those circuits.”

What's good news for enterprise customers who still live by TDM-based transmission is even better news for the carriers supplying them with DS-3, T-1 and other traditional leased lines. They can continue to get a return on existing infrastructure — and support a new round of revenue-generating services — without speeding up their IP migrations as the only way of providing quality Ethernet services.

That's exactly what the contributors to the ITU standard had in mind in 2002, just after they completed the standard for virtual concatenation in Sonet/SDH architectures. “The genesis for the circuit-bonding standard was really in optical network virtual concatenation,” said Jim Tindall, vice president of market and customer support at Ceterus Networks, which in September 2004 became the first vendor to announce an architectural solution based on the ITU circuit-bonding standard.

“There was a need in the Sonet world to create a way to support different-sized payloads, but after that initial work was done, there was a question of, what about the rest of the network — all those DS-3s and T-1s that are still out there?” he asked. “Ethernet is the biggest market driver out there, but these PDH circuits do not generate enough bandwidth. We needed a way to create more.”

Enter G.7043 and its related specifications. G.7043 states how to aggregate multiple circuits into one virtual pipe, while G.8040 and G.7041 specify the framing procedure for encapsulating specific transmission payloads. Also, G.7042 addresses how to grow or reduce the size of the bonded pipe by adding or subtracting a specific number of circuits.

“G.7043 is basically the envelope, and the other standards address what's in it, including how to seamlessly add and subtract from that pipe,” Tindall said. “In the Sonet world, there is a management overhead that makes a lot of that easier, but PDH networks don't have that, so we have to adapt something.”

The result, in addition to helping carriers create new service revenue and get more return on previous investments, also is something that might combine the best attributes of two different technology worlds. Telcos can support Ethernet, which has become the de facto access technology in most corporate enterprises, but they also can do so by leaning on the quality of service characteristics inherent to TDM architectures. Finding a way to do TDM on traditional circuits also means that they don't need to invest in newer methods of circuit activation, management and troubleshooting, said Dave Stehlin, president and CEO of Ceterus Networks.

“Overall, there's an operational and access cost reduction,” Stehlin said. That's because the bonding of circuits means fewer physical circuits are necessary to support multiple customer sites.

Circuit bonding can be used for Ethernet delivery to individual customer sites, but it also could be used to provide Ethernet bridging between multiple sites owned by the same customer, according to Karthik Muthukrishnan, director of strategy and port planning for Lucent Technologies' data access group.

“If you are in a metropolitan area with multiple customers sites, you can use your existing T-1s to bridge Ethernet into the other sites, rather than using more expensive routers for the job,” Muthukrishnan said.

Supporting Ethernet services in a TDM network environment is the primary application for circuit bonding, but vendors such as Ceterus and Lucent also believe that circuit bonding could provide a solution to the pressing issue of cellular network backhaul. There are a wide variety of technologies and approaches — both wireline and wireless — for providing cellular backhaul between cellular towers and mobile switching centers, but the method most widely employed by cellular carriers is to lease lines, usually T-1s, from ILECs.

However, cellular carriers are upgrading their networks to 3G at a quick pace, and customers are signing up in higher volumes for the data, video and other content services that take advantage of all that 3G bandwidth. Meanwhile, the growth in usage of those high-bandwidth services could potentially overload the capacity of those T-1s being used for backhaul, creating a bottleneck and resulting in poor service quality. Most cellular carriers are being increasingly careful in their network planning to avoid such a dire situation, and circuit bonding may be one of the most effective and cost-efficient ways of addressing it.

“There is a real trend in mobility, with higher-speed services and requiring Ethernet-like backhaul,” Muthukrishnan said. “The carriers have all those T-1 and E-1 leased lines that they don't want to rip out, but they may want to do backhaul by transporting Ethernet frames all the way from the base station. A lot of them are still using ATM frames, but that might not be a technically expedient solution.”

Stehlin added, “Providing Ethernet and linking together Ethernet LANs is the biggest application for this technology, but tower backhaul absolutely will take off down the road with all the announcements from carriers [about upgrades to CDMA 1X EV-DO and other 3G technologies]. Carriers will have to Ethernet-enable those towers.”

Despite the obvious applications and benefits of circuit bonding, not many carriers and vendors have rushed to celebrate it publicly since the standard was approved last year. In the vendor community, Ceterus and Overture Networks have launched products, while Lucent offers it as a capability.

“Circuit bonding is an idea that has been around for a long time — we can bond other kinds of circuits, not just T-1s,” Muthukrishnan said.

Because the general concept of circuit bonding has been around for a while, there could be a little bit of confusion about the standard and the fact that it actually addresses circuit bonding of PDH signals. Among other bonding techniques, there is DSL bonding, inverse multiplexing over ATM (IMA) and multi-link point-to-point protocol bonding.

But, Tindall, of Ceterus, said PDH circuit bonding best addresses Ethernet. “ATM bonding through IMA has been around for a while, and it works, but it's not really efficient for Ethernet,” he said. “It's also expensive because at this point, there is no reason for carriers to be expanding their ATM infrastructures.”

Two other vendors contacted by Telephony claimed to be aware of PDH circuit bonding and “interested” in working with it but declined the opportunity to discuss the technology further.

“My guess with most of the big vendors is that there is not a whole lot cooking there regarding circuit bonding,” said IDC's Perrin. “The last few years, they've had to cut back so much on research and development that with something new like circuit bonding, it's a solution the carriers may end up looking to the smaller vendors for.”

Perrin added that he believes vendors providing solutions based on the EFM standard would benefit further from also working with circuit-bonding technology in some respect.

Stehlin said Ceterus, which possesses patents related to circuit bonding, produced its own FPGA silicon because there was not a chipset vendor ready to deliver the silicon. He thinks, however, that it is only a matter of time before vendors arrive on the scene.

“It turns out that this is pretty hard to do because to have a circuit-bonding solution, you really need to support all parts of the ITU standard,” Stehlin said. “The chips aren't off the shelf yet, but all the big vendors have plans to do this.”

Rather than fearing more competition, Ceterus is welcoming it because telcos have a long-standing preference of wanting to see proof of multi-vendor interoperability before they deploy new solutions.

“We have been talking to carriers who want to do interoperability testing with us by the end of the year,” Stehlin said. “That's when we think the market will begin to turn.”

Meanwhile, carriers that watched the development of the circuit-bonding standard closely may just be getting to the point of starting trials, although they may remain tight-lipped about circuit bonding until they put it into practice commercially. Both Stehlin and Perrin also believe carriers might have other pressing strategic issues on their minds.

“There's a lot of M&A out there, and these guys are just trying to figure out where they will be when the dust settles,” Stehlin said.

“It's pretty surprising,” Perrin added. “Circuit bonding is something that should have strong appeal.”