AT&T Maven: 'Clean Up Your Networks'

NEW YORK -- Today’s metro networks are a mess.

That was the conclusion drawn by Robert Doverspike, of AT&T Corp. (NYSE: T), in a keynote speech presented at Light Reading's "Future of Sonet" conference at the W Hotel in Manhattan today. Pointing to a network topology slide of Gordian complexity, Doverspike, who is Division Manager for AT&T's Transport Network Evolution Research Group, told the audience: "My co-workers told me that I shouldn't share this slide with a Light Reading audience, but that's the whole point: The network we have today is a mess."

The problem is caused by having too many transport elements handling TDM muxing and demuxing and inefficient TDM-channeled handoffs to packet switches and routers, Doverspike told a packed audience of nearly 200 service providers and carriers.

“That’s what's inefficient in this network,” he said. "You have to do too many up and down transports, like a roller coaster."

Doverspike pointed to new packet-aware transport networks as a possible solution. "Let’s do this once, let’s not do all up and down two, three, and four times,” he added to nods and generally appreciative murmurings from a receptive crowd.

Packet-aware transport networks will transport legacy and emerging packet services from a metro network into the core via a unified infrastructure, according to Doverspike. These networks will incorporate Ethernet switching into metro transport platforms and use so-called "pseudo-wires" to multiplex packet traffic into Sonet virtual pipes.

Doverspike said transporting packet and TDM services over this sort of converged infrastructure, so the muxing and demuxing only have to be done once, could clean up the mess. He pointed to PWE3 (pseudo-wire emulation edge-to-edge) and Gigabit Ethernet-driven MSE (multiservice edge) boxes as keys to the desired convergence.

Today, PWE3 is used to emulate Frame Relay and Asynchronous Transfer Mode (ATM) over a Multiprotocol Label Switching (MPLS) network. Doverspike's suggestion that they be used to clean up the transport layer is new, according to Heavy Reading chief analyst Scott Clavenna, the conference moderator. “That’s the first time I heard a big carrier go out and talk about using pseudo-wire in relationship to a transport network,” he said. "Wow!"

Along with using pseudo-wires to create virtual circuits and MSEs, other packet-aware transport network technologies could include Resilient Packet Ring Technology and idle packet suppression (IPS).

Doverspike’s keynote hinted that AT&T will move to Gigabit Ethernet using RPR where possible, for price reasons. "Everybody tells us this trend [Ethernet being cheaper than Sonet] will change, but we don’t see it. As the price of Sonet comes down, the price of GigE comes down with it.”

Clavenna estimated that service providers won’t be deploying packet-aware transport networks before 2005, but Doverspike points out that they already run successfully in AT&T labs.

"The biggest problem is the lack of capital,” Doverspike said. “If we were in a growth period, people would just put this in. But the biggest driver of this is Ethernet, because of price. It’s going to start mostly with point-to-point Ethernet virtual circuits.”

— Dave Raffo, Senior Editor, Byte and Switch

rocksolid 12/5/2012 | 1:46:07 AM
re: AT&T Maven: 'Clean Up Your Networks' AT&T is complaining their network is too complex? Since when did AT&T ever take advantage of the latest technology.

For anybody on these posts that has ever tried selling ANYTHING into AT&T they will know that it is a 2-3 year sales cycle, MINIMUM, and these guys are NEVER the lead customer for anything.

And then they complain they have a complex network.

Maybe if they took advantage of technologies that have already been around for the past 2 years they wouldn't have the same problems.
zurg 12/5/2012 | 1:46:04 AM
re: AT&T Maven: 'Clean Up Your Networks' In the previous decade it was ATM that was supposed to "clean up" the networks. Everything was supposed to be transported over ATM Virtual Circuits: TDM, packets, voice, video, whatever....

A lot of people burned their hands (naively?) believing that this would actually happen. Well it didn't. ATM became just another layer in the ever growing deposits of telecom gear.

Now in this decade it's supposed to be Ethernet that will carry everything and replace all that there was ever been? Really?
Sign In