Optical networking equipment provides the bandwidth that carriers need to build high-speed networks. But to make money from those networks, they need something else: service-provisioning products -- specifically, the hardware and software that allows carriers to manage their networks, and automatically allocate capacity, track usage, and provide stats for billing.

Small wonder that service provisioning is fast becoming one of the hottest topics in optical networking – as evidenced at last week’s Supercomm trade show when four vendors made provisioning announcements: Tenor Networks Inc. http://www.tenornetworks.com, ONI Systems Corp. http://www.oni.com , Dynarc Inc. http://www.dynarc.com, and Alcatel SA http://www.alcatel.com.

A closer analysis of last week’s service provisioning announcements shows that beneath the interoperability demos, cutesy GUIs, and bravura sales pitches there are some razor wire-covered hurdles to surmount before these products deliver the goods.

The main problem is that there's no consensus on just how optical provisioning should be implemented. Right now, most vendors are using proprietary solutions -- including ones based on adaptations of MPLS (multi-protocol label switching)for optical networks. Work is under way to standardize this new version of MPLS, but it's no where near complete.

Experts also warn against making assumptions about touted MPLS support. "There's confusion about what it means to use MPLS," says Jim Lawrence, program director at Stratecast Partners http://www.stratecast.com, a consultancy. There will be key differences, he says, between MPLS that is used to provision IP networks in general and the version of MPLS that's being designed by groups like ODSI to work with special signaling protocols to configure optical connections. "The specs will use the same language, but there will be differences. English is spoken in the U.K. and the U.S., but there are fundamental differences in the way it is spoken" (see Crunch Time for Signaling Standard ).

Some even question how well MPLS will serve as an uber-protocol for provisioning. Vendors developing MPLS are "a bunch of lemmings," according to Stuart Barnes, director of engineering at Lilotron, a U.K. startup developing an all-optical switch. In his view, they are ignoring some harsh realities of carrier networks--such as the fact that incumbents have massive amounts of legacy gear already in place that can't be easily upgraded to support MPLS.

Others point out that there are some parts of carrier networks where MPLS may not be a fit at all. “MPLS will certainly be a big deal in the core. But it won’t be used all the way out to the metro edge of the network. That’s B.S!” exclaims Carl Russo, group vice president of the optical networking group at Cisco. Russo says MPLS will work well on backbone networks because they are built on a mesh topology -- the environment MPLS is designed for. Conversely, he says, MPLS is a poor fit for metro networks “…because they are physically built in rings, and will be for some time to come.”

Here’s the 411 on who announced what last week:

All of the vendors say they'll support MPLS-based standards over time from groups such as the IETF, Optical Internetworking Forum (OIF), http://www.oiforum.com), and the Optical Domain Service Interconnect (ODSI) coalition.

And none of them are willing to wait and see what comes of these efforts before heading to market. They have good reason: MPLS-based provisioning standards are far from finished. Initial specs from the OIF and ODSI aren't expected until later this year, with prototypes following in 2001. There are questions about how the groups fashioning the specs will work together.

-- by Mary Jander, senior editor, and Stephen Saunders, US editor, Light Reading http://www.lightreading.com