VPLS: money for nothing: it may be a fine protocol for upstarts, but VPLS is still not ready to unseat frame relay

America's Network,  June 15, 2004  by Dan Sweeney

• E-mail
• Print
• Link

"The great thing about VPLS is that it's essentially free to the service provider," says Matt Kolon, senior technical marketing manager for Juniper Networks, a prime mover behind the VPLS standard. "There's not even a software upgrade involved. Just enable the feature on our routers and you're in business."

But just what business are you in, one might ask, and there Kolon and other advocates of the emerging standard are not in harmony.

"I think that VPLS becomes the replacement technology for frame relay," says Eric Barrett, director of product management for Texas-based Masergy, an independent data communications services provider.

Kolon disagrees. "Maybe in the very long term, but right now I feel the two are complementary. A lot of people will want to retain frame access but have a dedicated layer two transport."

Lindsay Newell, director of product marketing for Alcatel, takes a different perspective. "VPLS shouldn't be considered in isolation. It's part of a portfolio of new MPLS-enabled service offerings. It's not one solution that's perfect for everyone. In some sense, it's part of an effort to redefine layer two Ethernet services, although it's a lot more than that."

SERVICE PROVIDER SCRUTINY

So what is the service provider to make of all this, confronted with the allure of money for nothing and uncertainty over the scope, capabilities, and destiny of the new protocol?

In the U.S. most service providers are clearly still struggling with such questions. VPLS has garnered considerable coverage in the trade press, and is clearly a topic of discussion in the industry at large. However, we've identified only three service providers, Masergy, SureWest (Fresno, Calif.), and MetNet (San Francisco), all independents, who actually offer commercial VPLS services, though representatives of SBC and BellSouth indicate that their companies are studying the market prospects of VPLS closely.

VPLS has gained the support of many of the equipment vendors that matter most, like Cisco, Juniper, Nortel, and Alcatel, but after two years of industry working group discussions, it isn't a real standard. With pre-standards commercial deployments of two years standing as well, it isn't a runaway success in the marketplace. More significantly, it is not without competition, namely, the well-entrenched IP VPN protocol using RFC2547 for signaling which is enjoying a solid success in the market today and which provides similar functionality to VPLS.

METRO ETHERNET REVISITED?

Several respondents suggested that VPLS can only be fully understood in the context of the overall metro Ethernet service model. Given the current revival of interest in that model, VPLS (which is unquestionably an extension of the model) might appear to be well positioned to enjoy rapid acceptance. Or is this in fact the case?

There is no question that metro Ethernet service offerings, after the failure of so many of them at the beginning of the decade, are again in market contention. Indeed, VPLS and other value-adds based on new protocols may be behind the resurrection of metro Ethernet.

Still, one must keep in mind that metro Ethernet revival has not resulted in that many Ethernet ports being activated in service provider central offices. And that, according to Massergy's Jim Brunetti, director of IP engineering, is holding back the adoption rate for VPLS and other value added services predicated on an Ethernet interface.

"It's possible to do VPLS with serial connections," Brunetti explains, "particularly if you use BGP. We do a lot of VPLS over T1, and if we didn't have that capability, we'd lose a lot of customers. But there's no question that a pure Ethernet platform is preferable."

Ron Young, the president of the startup service provider MetNet in San Francisco, amplifies on this point. "The presence of so much legacy equipment in incumbent networks is what is slowing down the adoption rate of VPLS. The benefits for the subscriber are obvious. For us it's not a problem because we aren't tied to legacy circuit networks, but others are."

But arguably it is a problem, though a problem for which partial solutions are already available. And it's a problem because, as Barrett insists,"VPLS potentially takes the metro out of metro Ethernet. It fundamentally changes the nature of the service, and, at the same time, it forces the service provider to adapt the service to whatever legacy access technology happens to be present wherever you want to set up a node."

Still, according to MPLS advocates, the benefits to both the service provider and the subscriber are such that VPLS is clearly in the ascendant.

RICH HISTORY

VPLS, like most new, new things in telecom, turns out to have a history. What it truly represents is one of several ways to implement VPNs over an MPLS network, and thus it may be regarded as but one prong of a larger MPLS onslaught.

VPNs, of course, have their own lengthy history, going back to the middle nineties and even before. Initial implementations involved nailed up circuits and were far from cost effective. Later offerings utilized frame relay and have come to represent the norm. Furthermore, frame VPNs are generally conceded to be highly successful both technically and commercially. So why the impetus for change?