Analysts see hope for U.S. long-haul
Over the past two years, carriers have simply plugged new transponders into existing systems when they've needed to increase capacity, and the next 12 months do not look any more promising, as carriers will continue to add line cards to meet traffic demand. A new study conducted by PointEast Research LLC (San Francisco, CA) reveals that there may be a glimmer of hope on the horizon, however. Of the 850 long-haul (LH) WDM carrier routes examined in the study, one in five is found to be greater than 75% utilised, requiring capacity upgrades within the year.
Between last November and January, PointEast analysts surveyed about 30 interexchange and regional carriers—representing two-thirds of the large service providers in tier one and two cities—to determine the overall utilisation of North American WDM networks. The research firm defines capacity utilisation as the ratio of the number of wavelengths that carriers have lit to the number they could light using equipment already installed.
The people at PointEast collected 1,161 data points, each of which represents a city-to-city link. They determined that the average utilisation among the 1,161 data points collected was 40.9%. The results were almost identical for 2.5 Gbits/sec (40.6%) and 10 Gbits/sec (41.2%). The average route has a capacity of 70 wavelengths or 35 channels, with just 28 wavelengths/14 channels lit.
Ostensibly, this data paints a less-than-rosy picture; there appears to be a large amount of unused WDM capacity on today's networks, and carriers should be able to meet demand using existing equipment. That said, traffic continues to grow—particularly on select city-to-city routes—and carriers are going to need some place to put it.
Of the 850 carrier routes studied, 321 routes had 10 or fewer unlit channels remaining, which will require "selective overbuilding of routes," say analysts. (Routes where a carrier had multiple systems in parallel were treated as a single system; thus, the 1,161 system routes were consolidated to 850 unique carrier routes.) Carriers are likely to add capacity to these select geographic areas within the next 12 months. Among those most heavily used are San Francisco, CA, to Salt Lake City, UT; Washington, DC to Philadelphia, PA; Chicago, IL, to Cleveland, OH; Dallas, TX, to Kansas City, MO; and Cleveland to Buffalo, NY.
"There's no magic number where once [carriers] hit that number, they go out and upgrade," admits Brian Van Steen, principle analyst at PointEast. "But a number of service providers are looking at deploying overlay systems and networks and are going through the process of evaluating different technologies and vendors right now." PointEast confirms that RFPs [requests for proposal] have been issued—even some for ultra-LH—with possible deployments beginning as early as the end of this year or the beginning of 2004. But are these select regional opportunities enough to trigger optimism among today's vendors?
"We think it's enough to be cautiously optimistic," asserts Doug Ranahan, director of business development at startup Ceyba (Ottawa, ON). "I'm reading this as quite good news in the sense that there's been virtually no new system deployments in North America over the last 24 months. What this study seems to indicate is that there are pressure points in the network that are sure to need new equipment over the next 12-18 months."
While established players Nortel Networks and Ciena accounted for most of the systems currently deployed, startups should not feel completely out of the running. Most of today's systems were deployed between 1999 and 2001, using the most advanced technology available at the time. As a result, they tend to be difficult to provision and manage and therefore expensive to operate and maintain. When evaluating systems for possible route upgrades, carriers will be looking for systems that are easier to operate, manage, and monitor. They will also be looking for systems that promise lower capital and operational expenditures. And though carriers are known to favour incumbents, "the startups typically have the more innovative solutions, the newer technologies, the new way of doing things," says Van Steen.
New system builds are not yet in the immediate future, but they will happen eventually—and they may not be tied strictly to capacity decisions, say analysts. "The goal is to have the lowest cost per capacity per distance," explains the report. "If deploying a new system can reduce the commodity cost by 50% or more, at some point a service provider will decide that it is more cost advantageous to deploy a new system even though the existing system has not been fully utilised." ..