Collapse pending

Service providers previously responded to user demand for new data services by adding new data networks. First came private line, then frame relay, then ATM and finally Internet Protocol and Gigabit Ethernet. Each service had its own infrastructure, its own network management and, in many cases, its own internal organization within the service provider.

This may be the year when public data networks begin to collapse.

Believe it or not, that’s a good thing.

Service providers previously responded to user demand for new data services by adding new data networks. First came private line, then frame relay, then ATM and finally IP and Gigabit Ethernet. Each service had its own infrastructure, its own network management and, in many cases, its own internal organization within the service provider.

This approach allowed service providers to bring new services to market relatively quickly – at least by a telco’s clock – without threatening the reliability of existing services. But by continually layering network technologies, service providers were also building in cost inefficiencies that would ultimately weigh down their business model.

For the last decade, the public network community has been abuzz with the prospect of collapsing all those layers of data networks onto one backbone network, with one set of operations supporting multiple services (This was ATM’s original purpose). While a single network sounds ideal, it is also complex – compare it to combining all the closets of your house into a single, multi-service closet. Now try to prioritize what stays, what goes, who gets priority access, how things are best arranged for multiple users, how many pairs of shoes each user is allowed to keep, etc.

Early attempts at creating a single multi-service network focused on building big iron at the core but that approach didn’t solve the really messy problem of handling multiple forms of access at the edge, where customers first encounter the public network.

Over the past two years, a new kind of product emerged – the multiservice edge device, a combination of switching and routing packaged to be cost-effective at the edge of a carrier network. Simply put, these MSE systems take in all manner of services from the customer side – frame relay, ATM, Gig-E, IP-VPN – and combine them onto an IP Multi-protocol Label Switched (MPLS) backbone for transport through the core of an IP network.

The idea is to use IP-MPLS to provide the guaranteed quality-of-service that ATM and frame relay have been providing for years now, while giving service providers a way to make the transition from multiple layered data networks to a single backbone.

That transition looks something like this: Install MSE device and use it to support growth in new and existing services. For example, when Business X gets tired of creating multiple point-to-point networks to tie every new location into the corporate network, its service provider is right there with a multi-point service offering that links the new site to the corporate headquarters and thus to the rest of the network. Meanwhile, other locations continue to chug along with existing services, as long as they are working.

Of course, nothing involving public network technology in transition is ever as easy as it sounds. This transition is no different. Here’s what to expect over the next 12 to 24 months:

1. Eight different vendors are now vying for the MSE space including the 9000-lb data gorilla, Cisco, with no less than three product lines, as well as its upstart competitor Juniper (two products), public network behemoths Alcatel, Nortel and Tellabs (each with brand new products); newbie Laurel Networks, which defined the space early on; as well as Redback Networks, a midsized player, and start-up Hammerhead Systems. The hype meter will soar as these players try to turn their product marketing strategies into network requirements.

2. Service providers will move, at their own pace, to use this new gear. Verizon is already out there, deploying a combination of Juniper and Cisco in its new broadband network. Equant is trialing Nortel’s new system. Laurel has a number of smaller deployments. The RFPs are out there, almost across the board, but that doesn’t mean everyone is rushing into volume deployment.

3. Large enterprise customers can play a significant role in how quickly their service providers move. The competitive landscape is such that service providers will be highly sensitive to what their customers need, even if they can’t turn on a dime to meet that need. Demand for IP-VPNs will play a significant role in advancing the multi-service agenda.

4. There is finally some indication of what the next generation of data services will look like. Integration of voice and data onto a multi-service network offers some real opportunities for follow-me e-mail and voice mail, customer portals for self-provisioning and many of the other things that, to date, exist only in PowerPoint presentations.

So while we hate to call any year “The Year of….” – this year is looking pretty good for multi-service edge products.