Sycamore Faces "Moment of Truth"
If plenty of vendors take the plunge, Sycamore can justifiably claim that its efforts to control the direction of some crucial standards have borne fruit, and that this will probably result in it gaining an important advantage in the market for optical switches.
On the other hand, if the interoperability tests don’t attract many participants, it will be hard for Sycamore to continue arguing that it’s business as usual at the ODSI coalition. More importantly, it will have lost the initiative and will be faced with having to catch up with vendors who’ve already charged ahead on a different course.
Right now, the latter seems more likely. Sycamore lost an important battle on August 17th when an ODSI proposal for a key signaling standard in optical networks was booted out by an overwhelming majority of vendors at a meeting of the Optical Internetworking Forum (OIF), another industry forum that’s higher up the standards-making food chain than the ODSI.
At the time, Sycamore said the rejection was “little more than an annoyance” and had resulted from a failure to get its message across (see Sycamore in Standards Setback). Since then, however, a different story has emerged. It turns out that other vendors had patched up their differences and put their combined weight behind an alternative proposal long before the OIF meeting, leaving Sycamore whistling in the wind.
The jury is still out on how this happened. One theory is that vendors deliberately ganged up on Sycamore to try and trip it up -- a sign of how worried they're getting about Sycamore’s ability to catch up and possibly overtake them (see Sycamore Ships Its Optical Switch). Another theory is that Sycamore has had its head in the sand. It failed to realize that the parade had moved on and that the time had come to jump onto a new standards bandwagon.
All of this started a couple of years ago when the Internet Engineering Task Force (IETF) and other standards bodies started work on defining signaling standards that would enable user equipment such as routers to automatically set up and tear down connections across optical backbones. The general idea was to extend MPLS (multiprotocol label switching), an evolving standard for setting up temporary logical connections over packet-based networks. The new standard became known as MPLambdaS.
The big argument at that stage was over what needed to be standardized. One bunch of vendors, led by Cisco Systems Inc. (Nasdaq: CSCO) and Juniper Networks Inc. (Nasdaq: JNPR), wanted the standards to embrace everything -- what goes on inside operator’s backbones as well as what goes on at the periphery, where user equipment is connected. No prizes for guessing why: The “peer-to-peer” approach encourages carriers to buy expensive routers for their backbones rather than lower-cost optical switches.
Not surprisingly, the vendors developing optical switches took the opposite view. They wanted to limit the scope of the initial standards to the periphery of carrier networks, so they could sell low-cost proprietary solutions for use in the optical backbones themselves. This approach is known as “UNI” (for user network interface), and the proprietary backbones are often referred to as “clouds”, because you can’t see what’s going on inside them, as you can with the peer-to-peer router networks used in today’s Internet.
The main advantage of the UNI approach is that it kick-starts commercial deployment and enables carriers to preserve commercial secrets and keep a tight grip on security. “At least for now, the carriers are pretty happy with a closed domain, a cloud,” says Joe Berthold, chairman of the OIF’s technical committee and vice president of network architecture at Ciena Corp. (Nasdaq: CIEN).
In the long run, however, the peer-to-peer approach promises to encourage competition, by opening up equipment markets and enabling carriers to use each other’s infrastructure. “The more open systems are deployed, the more vendors and the more carriers there’ll be,” says Khrishna Bala, chairman of the OIF committee handling signaling.
Most people think that both approaches will coexist in carrier networks sooner or later.
Here’s where things start to get complicated. At the beginning of this year, Sycamore opened a third front in this battle by proposing a fast-track way of developing a UNI standard and setting up the ODSI coalition to take it further (see Third Front Opens on Standards War). Part of this includes a different signaling protocol, one that shortens development times by enabling vendors to reuse existing TCP/IP software stacks, according to Amy Copley, a senior product manager at Sycamore.
This proposal split the UNI-promoters into two camps. The bigger vendors stuck to the OIF while the smaller players initially backed Sycamore and the ODSI. However, the big carriers weren't happy with the ODSI's approach, according to John Drake, chief network architect at Calient Networks and a key developer of MPLambdaS standards. “The people that want MPLS and MPLambdaS are the people wanting to do traffic engineering,” he says. Sycamore’s TCP/IP-based signaling doesn’t cut it on this count, he adds.
In any case, Sycamore’s initiative has became sidelined by the settling of the bigger battle over whether the inside of carrier backbones should be standardized. A large group of vendors are now backing a scheme that brings the UNI and peer-to-peer models under a single umbrella so that standards-based equipment can be used for both environments. “This means that talking about peer-to-peer versus overlay is a red herring or a blue whale. If you support the standards, you don’t care,” says Calient’s Drake. The only ones who do care are vendors like Sycamore that have charged ahead in advance of standards and now need to protect proprietary product lines, he adds.
Drake’s view is based on MPLambdaS standards being deployed really quickly. He maintains that the standards “are basically done,” but others take a more conservative view. “It’s going to take a while -- one or two years -- for the IETF to complete its work” says the OIF’s Berthold. “There are still a lot of battles going on with MPLS specifications,” says Jeffry A. Kiel, Sycamore’s VP of product marketing. The process of implementing the standards in equipment and getting it tested and deployed in carrier networks is likely to add further lengthy delays.
Still, getting most vendors to agree to a common set of standards is quite an achievement. “It’s been an amazing ride for the past five months,” says the OIF’s Bala, who is also CTO of Tellium Inc., the vendor that proposed the umbrella solution. Vendors backing this approach include Calient, Ciena, Cisco, Juniper, Lucent Technologies Inc. (NYSE: LU, Nayna Networks Inc., Nortel Networks Corp. (NYSE/Toronto: NT), Sorrento Networks Inc., Tellium, and Zaffire Inc., according to Bala. He also says that the scheme is being backed by "at least ten" major carriers.
Surprisingly, Sycamore professes total ignorance of these developments. Bala says he approached Rick Barry, a founder and CTO of Sycamore, on a number of occasions but couldn’t persuade him to get with the program. Sycamore denies these conversations ever happened. In Bala’s view, the vote at the August 17th OIF meeting merely confirmed what everybody else knew -- that Sycamore was out of step with the rest of the industry and wasn’t listening to what people said.
The real lie of the land will become clearer when the ODSI starts testing interoperability of equipment based on its proposed specs. As noted, vendors have to commit by next Saturday. The tests themselves are scheduled to start on November 6.
-- Peter Heywood, international editor, Light Reading http://www.lightreading.com