Optical/IP Networks

The Softer Side of Convergence

WASHINGTON, D.C. -- It's the software, stupid.

That's the message coming from carrier network architects on hand at the Light Reading Live Convergence 2004 conference, held here Monday.

Nearly 100 brave souls ventured through a snowstorm to talk shop on topics such as SIP, MPLS, and OAM. Attending service providers included:

Some equipment vendors and enterprise networking customers were also on hand.

At issue was how to integrate -- or converge -- newer IP-based technologies with the legacy systems inside complicated carrier networks with thousands of nodes. Doing so will cut costs and allow carriers to deploy new services, but several technical challenges remain. (See Setting a Course for Convergence and Carriers Say VOIP & SIP Are Hot.)

At the top of the list of carrier gripes are software bugs and disparate management systems.

“I’m against regulation, but maybe there’s some new [management] standards that can come out in the industry," said Clayton Lockhart, vice president of architecture and strategic operations planning with AT&T. "I’m tired of taking on new vendors and taking out their OS bugs. How and why do I want to get off the platform I'm on?"

Lockhart, making one of the two carrier keynote addresses, said his company is shying away from new equipment suppliers for this very reason -- and will continue to require potentially new partners, including startups, to work with their existing vendors.

"We need some way in which I’m not always relying on one vendor’s EMS [element management system]," says Lockhart. "I need a way to not proliferate the EMS trash."

Other participants at the LR Live conference concurred, pointing out that proliferating proprietary management systems are wearing thin on carrier patience.

"We've got 100,000 boxes, which are a management nightmare," said Phil Holmes, CTO of BTexact Technologies, the other keynote speaker.

Holmes noted that it's the various CLIs (customer line interfaces) for different equipment that cause problems. Moving to new platforms requires them to retrain employees and learn new management systems. "We need a consistency of commands across the industry."

The answer, he believes, lies in better standards and management protocols, which aren't adequately addressed in the current round of Multiprotocol Label Switching (MPLS) standards at the Internet Engineering Task Force (IETF). Citing an an example, he said the IETF currently has no standards for OAM (operations and maintenance).

Heavy Reading chief technologist Geoff Bennett, the moderator of the conference, also focused on the need for better management tools. Bennett observed that existing standards for fault-tolerance, such as NEBS, largely address hardware and include few provisions for operating systems.

"Maybe we need new standards for software performance," said Bennett. "They have this in the military."

Over time, the problem may be getting worse, said Bennett, because newer packet-based equipment often includes code that's pre-standard -- and even runs it live in carrier networks. “These days customers are actually running draft implementations… It’s frightening."

Bennett pointed out that service providers are equally to blame for this problem -- they should become more involved in the standards process and pipe up when something isn't right. "You've got to show up and voice your opinion."

Some equipment provider representatives agreed that carriers themselves need to be more aggressive in moving foward.

"It's like a weight-loss program -- in order to change things you've got to be ready to change" said Sam Halabi, director of marketing for advanced data products with Tellabs Inc. (Nasdaq: TLAB; Frankfurt: BTLA). "Also, if I had a message for carriers, it's that the perfect solution is not going to come until my box gets into your network... It's the chicken-and-egg problem."

— R. Scott Raynovich, US Editor, Light Reading

Sign In