Carl Weinschenk spoke with David Hold, senior analyst for Network Services for Current Analysis. The firm released a report earlier this month, "Next-Generation WAN Services: Enterprise Migration Plans and Trends."
Weinschenk: What did the study look at, and what did you find?
Hold: What we were looking at is enterprise network migration plans and to what extent they are using legacy technologies, ATM and frame relay. We also looked at their plans to migrate from the old to new stuff, from legacy networks that are widely deployed on a global basis to the next-generation of IP VPNs and Ethernet services. We found that there still are significant numbers [of companies] using traditional network services, and a big percentage are planning to migrate to IP VPN or Ethernet within the next two years.
Weinschenk: Have you done this type of research before?
Hold: This was the first time we've done this kind of research. What we were really looking for was the percentage and timeframe. ATM and frame relay were the mainstays of networking tools of the 1990s. Since 2000 and before, IP VPNs and Ethernet have been competing with legacy technology. This survey was geared toward larger enterprise customers as opposed to small businesses. Two-thirds of the respondents spend more than $1 million a year in telecom and one-third more than $10 million a year. A large percentage of large customer networks have not migrated. It's simply difficult to migrate a large multinational. They consider their options very carefully over a long period of time.
Weinschenk: You point to a contradiction between the incumbent carriers and some of their big clients. Please explain.
Hold: The incumbent carriers, Verizon, AT&T and others with large installed bases, have a legacy technology challenge. All their investments are going to the new stuff, IP networks, MPLS networks and Ethernet. But they have this core of their largest customers. The larger the customer, the more likely they are to use legacy networks. Carriers are not investing in that, and would really like people to begin that migration.
Weinschenk: Sounds like a problem.
Hold: It's a manageable problem. There is a question of how you deal with it. One example is by announcing the end of life of the traditional network. They give customers a number of years to get off the service and to get to something newer. Verizon and AT&T so far are not willing to do that. Sprint has "end of lifed" its legacy traditional ATM and frame relay service. The other interesting point is that until 2006 most revenues at large carriers were still coming from legacy services. The industry is just now getting to the cross-over point where the next generation is showing larger revenue than the legacy networks. That's why this is a particular topic of interest, because that's just happening now.
Weinschenk: So what are some of the results of the research?
Hold: Seventy-five percent of frame relay users say they are migrating all or part of their services to IP VPNs in next two years. That's a pretty overwhelming percentage. IP VPNs clearly are favored as far as frame relay users go. Forty-six percent of frame relay users say they plan to migrate to carrier Ethernet in the next two years. There are a lot of overlaps, because we asked companies to list [all the technologies they] plan to migrate to. Companies are using Ethernet for access and in the metro area for facilities requiring very high levels of bandwidth. Branch offices more likely have slower requirements and are more comfortable switching to an IP VPNs.
Weinschenk: What is driving the growth of IP VPNs?
Hold: Two things are really driving the uptake of IP VPNs. One is the convergence of voice, video and data on the same network. The other one is the lower cost and lower cost-per-bit ratio. The lower-cost-per-bit is a better measurement. They may be spending more money but getting a lot more bang for the buck. If you go from a T1 frame relay to 10 megabit per second Ethernet, you may double the cost but you are getting six or seven times the bandwidth. One final driver for IP VPNs is the ability to create multipoint or meshed topologies.
Weinschenk: What is driving Ethernet growth?
Hold: For frame relay to Ethernet, the drivers are a little different. We asked people to rank the first, second and third reasons. One that really stands out is that they are looking for bandwidth. It's that simple. That was even stronger than convergence. Convergence was second, business application requirements was third. We see differences between frame relay and ATM users and what they are going to do. ATM users are more likely to select Ethernet than frame relay users. That makes sense. ATM is a high-bandwidth networking service, typically in the core of the network. Frame relay often is at branch offices. Most frame relay connections are T1s or below. So high-bandwidth requirements make for a natural migration to Ethernet.
Weinschenk: What is the timeframe on all this?
Hold: Eighty percent of ATM users will migrate [at least partly] to IP VPNs and 61 percent to carrier Ethernet over the next two years. That's a big difference from frame relay users. Forty-six percent of them plan to go to carrier Ethernet.
Weinschenk: Why the disparity in rollouts?
Hold: I think IP MPLS is pretty fully baked. Ethernet is still a work in progress. Ethernet is very well deployed in metro areas. It is increasingly used as an access technology, sometimes used in place of TDM-based private lines. When you get outside the metro area, Ethernet availability is spotty. The ability of carriers to deploy long haul comparable to ATM and frame relay still is somewhat limited. Ethernet is very strong in the metro and it's increasingly used as an access technology. The problem is in long-haul locations that are distributed among many cities on a national or global basis. Ethernet is in its early days on that.
Weinschenk: What about virtual private LAN (VPLS) services?
Hold: What our survey found is that VPLS is not widely deployed. There is a profound lack of knowledge about the service. The technology was not available from major tier 1 carriers until this year. If more large carriers start launching service, that could change. VPLS allows for an alternative migration, which today points to MPLS VPNs or Ethernet VPNs over distances and multiple locations. VPLS could make a very good replacement for ATM or frame relay networks.