Sprint touts net simplicity over MPLS

But analysts say MPLS is likely to play a key role in carrier networks, meaning a possible risk for Sprint.

Multi-protocol Label Switching has gotten a lot of attention from equipment vendors, carriers and the technology trade press. And that attention has been largely positive, with the exception of one notable source: Sprint.

Sprint officials insist their company is neither officially pro- nor anti-MPLS, but they say adding MPLS to Sprint's IP backbone would create unnecessary complexity.

While industry analysts say MPLS might not be necessary for pure IP networks, the technology could be useful as a transition from connection-oriented frame and ATM services to connectionless IP services. And if Sprint doesn't at least look into implementing MPLS in the future, the carrier could fall behind competitors.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

Multi-protocol Label Switching has gotten a lot of attention from equipment vendors, carriers and the technology trade press. And that attention has been largely positive, with the exception of one notable source: Sprint.

Sprint officials insist their company is neither officially pro- nor anti-MPLS, but they say adding MPLS to Sprint's IP backbone would create unnecessary complexity.

While industry analysts say MPLS might not be necessary for pure IP networks, the technology could be useful as a transition from connection-oriented frame and ATM services to connectionless IP services. And if Sprint doesn't at least look into implementing MPLS in the future, the carrier could fall behind competitors.

"No national long-distance carrier has any option but to make a commitment to MPLS," says Thomas Nolle, president of consultancy CIMI and a Network World columnist.

MPLS is an IETF standard that is designed to attach labels to traffic flows. The thinking behind MPLS is that it can be used to prioritize traffic and avoid network congestion or failures. Carriers such as AT&T, WorldCom and Cable & Wireless already have implemented MPLS in some form.

But Sprint insists that MPLS only adds unnecessary complexity to IP networks.

At this fall's NetWorld+Interop show in Atlanta, Barry Tishgart, Sprint's director of dedicated data services, described MPLS as a "solution looking for a problem."

The reason Sprint doesn't need MPLS is that the provider has overprovisioned its OC-192 SprintLink IP backbone network and doesn't hit more than 40% capacity on either of its two, physically diverse backbone paths, says Randy Ritter, vice president of product/portfolio management. Other carriers are deploying MPLS, Ritter says, to manage congestion on their networks. Sprint doesn't need MPLS, because there is no congestion.

"We feel like we've already deployed the network other carriers are moving toward," he says.

Sprint does some quality-of-service engineering for its IP customers. The provider uses Differentiated Services to establish up to five traffic queues between customer premises routers and Sprint points of presence in major metropolitan areas. But once the traffic hits the SprintLink backbone, it's all treated equally and routed at Layer 3.

"We believe we can handle the requirements customers are placing on us by managing from our edge router to the customer premises," says Fred Harris, Sprint's vice president of research, architecture and design. "The idea is to push the intelligence to the edge of the network to keep the core of the network as simple as possible. The core of the network is what needs to scale."

Steven Taylor, president of consultancy Distributed Networking Associates, publisher/editor in chief of Webtorials.com and a Network World columnist, says customers shouldn't really care about the technology behind their VPN service.

"It shouldn't matter as long as you're getting the service-level agreement you need," he says.

And Nolle agrees that MPLS doesn't add anything to pure IP networks. But, Nolle says, frame and ATM are becoming too expensive for carriers to deploy profitably, and by 2005 he expects frame and ATM to become unprofitable.