Multilink frame relay might gain fans

Getting the most out of existing network technologies is at the forefront of just about every network executive’s mind. For some, multilink frame relay might be the latest “old” service that fits the bill.

Getting the most out of existing network technologies is at the forefront of just about every network executive’s mind. For some, multilink frame relay might be the latest “old” service that fits the bill.

Multilink frame relay lets users bond multiple T-1, 1.544M bit/sec connections so they can incrementally increase bandwidth without migrating to a new technology or costly T-3, 45M bit/sec, ports.

While the service has been around since 1999 when the Frame Relay Forum issued its first implementation agreement, or specification, more users are expected to adopt the service over the next few years, according to analysts. The main drivers are the economy, a recently updated specification from the forum and the promise of standard-based services.

“The bad economy may be good for multilink frame,” says Steve Taylor, president of consulting firm Distributed Networking Associates and a Network World columnist. “Despite the bad economy, users are still dealing with traffic increases.” To keep up with network demand, existing frame users would rather keep their current infrastructure if they can upgrade to higher speeds economically, he says.

Other analysts agree.

“Frame relay is the most prevalent service out there after private-line services,” says Erin Dunne, research director at consulting firm Vertical Systems. “And it isn’t going away despite all of the hoopla about IP VPNs. Bandwidth needs are increasing, and multilink offers users a good alternative to T-3 service.”

Sprint will offer its frame customers a fully compliant multilink service based on the forum’s FRF 16.1 specification next year, says Larry DeNayer, manager of frame relay products.

Sprint has offered a proprietary service since 1998, but the standard-based version is expected to bring two primary benefits. Customers could deploy any FRF 16.1-compliant vendor device at their site, and it will allow Sprint to free up considerable network resources, DeNayer says.

The carrier offers a service that is based on Quick Eagle devices, which are required at every multilink location. The new service will eliminate that chokehold, letting customers use routers and devices they already own.

Sprint supports its proprietary service by deploying a multiplexer that sits in front of its Nortel Passport 15000 multiservice switches. That multiplexer uses up an entire slot on the Passport for every multilink customer using up to eight T-1s. The standard-based service will eliminate the need for the multiplexer and will let Sprint support up to 128 T-1s per slot, compared with eight with its proprietary version.

“That’s a significant difference,” DeNayer says. He says the upgrade will simplify network support and reduce costs that could trickle down to customers in the form of lower service rates.

AT&T and WorldCom traditional frame relay customers are not likely to see a standard-based service any time soon.

WorldCom is offering its IP customers a multilink T-1 service that is based on the forum’s FRF 16.1 specification, which was ratified in 1999. While it is an IP-only service, it runs over WorldCom’s UUNET network that uses frame relay at its core. While analysts think this is a smart move and gives multilink frame wider appeal, traditional frame users are still left with a proprietary offer.

WorldCom’s NxT-1 service is based on Quick Eagle technology, as well as Larscom gear.

AT&T customers do not have a multilink frame relay option. Frame customers who want to increase bandwidth beyond T-1 have two choices: They can opt for AT&T’s ATM Inverse Multiplexing service that lets users bond multiple T-1 ATM connections, or they can buy a dedicated T-3 port.

But a T-3 port can cost up to 10 times as much as a T-1 port, according to Sprint.

Both WorldCom and AT&T say they are “investigating” the need for a service based on FRF 16.1.

Senior Editor Tim Greene contributed to this story.