Nortel CTO mixed on divested products

Nortel has had to divest numerous product programs to cut expenses and return to profitability, but if CTO John Roese could have any of them back, he would take network management.

"The one [area] that I think the company may have made a mistake around . . . was maybe an overly aggressive divestiture of the [network] management assets," he said last week in an interview at Nortel's laboratories in Ottawa, Canada. Roese recently concluded his first three months on the job at Nortel.

"The reality is now we find ourselves in a position where network management is a build-up exercise for us," Roese said. "We're having to reinvest and we have at least a dozen different projects going on to rebuild and develop next-generation management technologies. We're starting a little bit behind where I'd like us to have been and maybe some of those assets would have helped us. I'm a big believer that the operational aspects of the network are almost as important, if not more important than the transport applications."

The divested management assets include operations, administration and maintenance (OA&M) code that stretched across Nortel's telephony and enterprise product portfolios. But the silver lining is that Nortel now gets to start writing those applications using a clean slate as telecom and enterprise infrastructures go through next generation transitions. Service-oriented architectures (SOA) in the enterprise and IP Multimedia Subsystem (IMS) frameworks for carrier networks require new disciplines for network management, he says.

"A stronger management play would have been nice to start with; the reality is there's some pretty significant reinvention in that place, too," he says. "Whether it be IMS . . . or SOA, we get a bit of a clean sheet of paper. So the good news is our management activities are around SOA and Web services and next generation concepts. But...I think probably the weakest link was the management frameworks that exist here."

Roese was one of the architects of Cabletron's heralded Spectrum management platform. He is also the named inventor on 16 granted and pending patents for policy-based networking, location-based networking, routing, switching and network management.

Good riddance to old access business

One Nortel business Roese was happy to see go was access. Nortel spun off its DSL business years ago, and a joint venture with Huawei on Gigabit Passive Optical Networks (GPON) never got off the ground.

GPON is being deployed by Verizon, AT&T and BellSouth as the access infrastructure for their multibillion dollar fiber to the home, curb and neighborhood buildouts to offer high speed Internet access and video service. Verizon has said GPON will support gigabit-per-second downstream and upstream speeds.

Page Break

"Broadband wireline access is right now such a churning space, it's in such a state of evolution that I'm glad we don't have an attachment to the DSL world or a particular type of PON technology," Roese says. "The customer hasn't decided yet. It maybe limits our ability to go after certain market segments in the carrier edge market right now, but as we start to plot that market what we're finding is there is a general feeling by the carriers that what they'd like to see is something that can scale far beyond existing PON or DSL technologies, and is operationally significantly simpler."

Nortel and Korean vendor LG Electronics, through their joint venture, are trialing a wavelength division multiplexing PON technology that offers some promise of that simplified scale to which Roese referred. Once the fiber plant is in place, the electronics to deploy something other than GPON is not complex, he says.

And as carriers look to eliminate those electronics to further reduce operational complexity - making the network even more passive than PON and placing more of the processes in the optical domain - it should be lower cost, higher speed and simpler to operate, Roese contends.

"Our vision is, as the market matures - as we start to find technologies that can actually scale not up to the tens of megabits per second, but up to gigabits or tens of gigabits per second to the endpoint - that would be a very interesting place to reassert ourselves," he says. "But in the meantime, when you have a choice of 100 different technologies of which no one is dominant and they're all just continuously oscillating, it's actually better to take a step back, make sure your interfaces are open, partner well, and let it sort itself out instead of trying to influence it towards the end state that you're after.

"If you bet your company on something like GPON, I'm not sure that would be a very wise bet if you're a large incumbent player," Roese says. "With the stroke of a standard you could be rendered obsolete and your technology could be essentially irrelevant."