How big, how soon?

During the next 12 to 18 months, communications carriers are going to have to make critical decisions about their network infrastructures. Triggered by the ascendancy of data traffic and stoked by technology breakthroughs such as wave division multiplexing and optical networking, all these tough decisions revolve around how best to converge voice and data traffic.

Ultimately, most carriers seem to be moving toward Internet protocol-based networking because it is already used in local networks at each end and it has less overhead than its main contender, asynchronous transfer mode. Eventually, IP will support quality of service and real-time traffic without requiring the time-consuming - and less scalable - process of setting up the virtual circuits associated with ATM.

Accurately anticipating how quickly the shift to IP will occur - and how encompassing it will be - could mean huge net gains for carriers, while inaccurate assumptions could mean huge net losses. Not surprisingly, service providers are commissioning survey after survey to help make these key decisions.

One area that's receiving a great deal of scrutiny is virtual private networking - a means of securely and reliably mixing multiple types of traffic from multiple users over a common IP path.

Two of the more bullish forecasts: Strategic Networks projects that 60% of current private line networks will convert to IP-based VPNs by 2003. The Yankee Group predicts that by that same date, 70% of all companies will use IP-based VPNs for up to 90% of their data communications needs.

The key question on which these forecasts hinge: What will be the compelling reason for customers to adopt VPNs?

Because VPNs cost less to deliver, service providers hope to lure customers with lower pricing. But first they have to convince customers that quality will not suffer.

Previously, carriers focused on finding a killer application for each new service. But when it comes to VPNs, Internet service providers seem obsessed with finding the killer service level agreement. How many milliseconds of latency? How much packet loss? How many days to installation?

The assumption is that customers won't mind putting their critical data on the Internet or another IP network if they are satisfied with performance - and they can save some money. Early adopters, though, may be looking for something other than a better price. VPNs can offer something other transmission options cannot: ubiquity.

Unlike other data networks, the Internet can be reached easily from just about anywhere a business customer might travel - and business travelers are itching to connect to corporate networks from the road and get all the capabilities they have sitting in their offices. To get an idea of how large this market is, just ask information technology managers how they define a VPN. Many of them will describe it as some form of secure remote access.

The Internet's ubiquity also may give VPNs special appeal to multinational corporations or to companies that want to connect with trading partners on an ad hoc basis. Other early VPN adopters will be small companies or branch offices that have some kind of data connection but are without Internet access. Why settle for just data when they could browse the Web and get e-mail too?

Interestingly, many of these hot VPN applications are just a piece of a larger organization's complete communications requirements. Many of these large organizations may be content with the way their other communications needs are being met - at least for now.

These circumstances actually may give telco-affiliated ISPs an edge - if they choose to use it. Sprint has the right idea with its SLA that maps to its conventional carrier services, such as ATM and its Internet offering.

Similarly, MCI WorldCom has installed frame relay-to-Internet gateways to give frame relay customers an easy way to connect to the UUNet Internet backbone, through which users can reach IP-connected trading partners or branch offices.

I wouldn't dare to predict how quickly VPNs will supplant conventional data connections, but I'm convinced they're coming on strong. Now is the time for telcos to leverage their ability to support IP and conventional data services to get a lead in this market.