Last night at the MPLS 2002 conference here, representatives from AT&T Research Labs, Cable & Wireless (NYSE: CWP), and WorldCom Inc. (OTC: WCOEQ) participated in a panel discussion about where MPLS is headed in their networks, how they plan to get there, and what they need from startups.
While it’s no surprise that all three carriers agreed that Multiprotocol Label Switching (MPLS) was in the future of their networks (given that they were speaking at an MPLS event) the carrier technologists surprisingly agree on a general strategy and the types of products they expect. The detailed loquacity of the carrier representatives also came as a surprise, since they typically aren't fond of public speaking (see MPLS vs ATM? Vendors Weigh In).
So what did they say?
- Network Convergence: Yes, it’s been hyped for years, but these network architects say their companies are committed to MPLS and packet-based networks. They said MPLS promises to offer substantial cost savings, not just on a per-bit basis, but also through improved operational efficiency.
“The transport of ATM or Frame Relay in large networks is already efficient,” said Chris Chase, technology consultant at AT&T Research. “The hope with MPLS is that things will be more efficient from an operational standpoint. Maybe we can reduce the number of customer call centers or reduce the number of network operations centers running; retain customers through improved troubleshooting.”
- ATM Is Here to Stay: Every service provider agreed that there is too much legacy gear in their networks to justify replacing the old with the new.
“We don’t want to take well utilized gear, rip it out, and buy something new,” said David Garbin, vice president of network strategy for Cable & Wireless. “But we also don’t want to keep buying any more ATM gear.”
This means that the road to convergence is going to be a long one; and the poor economy hasn’t helped much, as these carriers are being pushed to stretch their current networks further than ever before.
“The days of deploying a new technology and expecting traffic to fill up the pipes are over,” added Garbin, during an interview after the panel discussion. “Demand for the network is still growing, so we’ll eventually be spending money on new gear -- and it won’t be ATM.”
- Migration Strategy: Keep old customers on the legacy network, while building a new MPLS network for new customers and new markets. Garbin emphasizes that these transition networks will likely be in place for years rather than months.
Exactly how this convergence is going to happen is still up for debate. AT&T's Chase presented several scenarios wherein overlay networks were suggested or a parallel network approach called “ships in the night” was taken, but he admitted that he wasn’t too happy with any of these architectures.
”We’re still debating these internally,” he said.
A whole slew of startups and established players have already built products designed to head in this direction. Chris Liljenstolpe, senior director of network technologies at Cable & Wireless, said that his company has already narrowed its short list of candidates in this product category to: Cisco Systems Inc. (Nasdaq: CSCO), Juniper Networks Inc. (Nasdaq: JNPR), Laurel Networks Inc., and Redback Networks Inc. (Nasdaq: RBAK). He said they expect to deploy the gear sometime next year.
But the carriers also said that one key piece of the puzzle is still missing in all these platforms: a resiliency signaling mechanism that can connect the old and new networks. ATM networks use PNNI signaling, while MPLS uses its own signaling to create label switched paths through the network and detect outages. These two signaling protocols don’t work together, which means there's no interoperation between them.
“Without a control plane signaling between the networks, we have to develop our own operations software to make it work,” said Chase.
Building a compatible, converged network is not an easy task, by any means. Liljenstolpe said part of the problem is that the standards bodies have not been working together toward a solution. The ATM Forum, which establishes PNNI standards, has been working independently from the Internet Engineering Task Force (IETF), which is developing the MPLS standards. But in the end, he said, it will come down to the vendors that must get the job done.
“When we tell a vendor this is what we want, they get all white in the face and run the other direction,” he said. “They say ‘Oh, we’ll get back to you on that.’ ”
But new gear isn’t the only thing these carriers are asking for. They also want more reliability in current and future products.
“ATM gear from the telco world is steeped in reliability,” says C&W's Garbin. “But the Internet gets its reliability from many unreliable devices. If you move mission-critical applications to these multiservice edge devices, you will need to address that issue. It’s a culture shock for the IP guys, I think.” — Marguerite Reardon, Senior Editor, Light Reading