Is There a Bandwidth Glut?
By Ken Branson
from the June 4, 2001 issue of Broadband Week
Yes, there is a bandwidth glut. And no, there isn't.
Here's the situation. There is a lot of glass in the ground, and a growing amount under the water, so the overall capacity is so vast that the term glut may not be out of line. But the amount of lit capacity remains quite small, reflecting the expense of actually using the capacity.
And then there is the metro market in North America and Europe, where, emphatically, there is not a glut. Quite the opposite, say industry insiders.
"I think where we're at right now is, companies have built these pan-European and pan-American networks, and the builders have sold strands to other providers, who are in the process of lighting them, and there is a large amount of capacity," says Donald Noonan, vice president-networks at Band-X, the bandwidth exchange. "Particularly in the U.S. this year, a lot of new nets have been built. There appears to be enough fiber in the ground to satisfy demand for the next five years."
However, fiber in the ground or under the water does not a glut make, according to Jerry Power, director of marketing/emerging markets at Alcatel.
"Providers install first a skeletal system," Powers says. "The cost of that type of equipment is all incurred as you equip it. If you look at the fibers, even though you have fibers in the ground, the electronics at the end are only partially turned on. So what ends up happening...the cost is mostly in the plug-ins, and they can defer that cost until they have a service demand."
The Big Question
"Is the demand there?" asks Noonan. "Are the apps for this bandwidth waiting in the wings? And is the customer going to be able to get service to a location that's useful to him? And those are the answers that will determine whether there is a 'glut'."
Many analysts believe, based on past experience, that applications appear to fill available bandwidth; as more is available, sophisticated apps that demand more bandwidth get developed. Then the bandwidth expands, and we begin to think about gluts again.
"We believe that this is an industry with a constant see-saw between undersupply and oversupply," says Ford Cavallari, global commerce and content practice leader for Adventis, a Boston-based consulting firm. "The industry is never going to hit it straight on."
Noonan is dubious. "There's all this bandwidth," he says. "Carriers have to light it. In the U.S. there are only about three dozen strands of fiber actually lit. I'm not talking about a fat, short ring in New York City. I'm talking about nationwide networks...I believe...through 2003, there are plans for an additional 570 strands to be deployed. The jump from having dark fiber to lighting it is a tremendous jump because of the cost involved. I've seen estimates from $500 billion to $1 trillion to light that fiber. Not all these networks will get built, and most won't get lit."
Demand Trends
With the economy slowing down, with Internet-intensive businesses failing, it might appear that demand is slowing, and that, therefore, the justification for lighting existing fiber might be a long time coming. But Chris MacFarland says his customers still are demanding more bandwidth. MacFarland is vice president of technology at Dallas-based CLEC Allegiance Telecom.
"Our business is retail, and our end-users are small to medium-sized enterprises that need eight to 12 business access lines and some sub-T1 Internet access," MacFarland says.
Despite all the hype about serving small and medium-sized business, MacFarland and his colleagues claim that market remains underserved, and that demand is pent up in many metro areas. MacFarland looks to technology to increase the capacity of existing fiber, in the metropolitan network, as much as in the long haul.
"It's not a mature industry, and a series of technical advances are making it more mature," he says. "The whole evolution, circuit to packet...advances at the core (of the network), and is slowly migrating to the edge. The core of network is more mature than the edge."
The technological advances of recent years--in particular, dense wavelength division multiplexing, or DWDM--have made it possible to get more information through less and less fiber, even as more glass is deployed.
"You know, the equipment currently being installed can deliver 40 10-gigabit wavelengths over a fiber pair," Noonan says. "And Nortel is shipping 160 40-gig wavelengths this summer. So, with the same amount of fiber, you can now transmit 6,400 gigabits vs. 400 gigabits. So, that's 16 times the capacity over the same amount of fiber. Even if one were to lay no more fiber, hardware tech advances would provide more capacity over the same fiber."
Where's The Supply?
But if there is a glut, Tony Thakur asks, why is buying long-haul capacity such a pain in the backside? Thakur is vice president of technology at CLEC Time Warner Telecom, which often buys long-haul capacity for its customers. Ask Thakur if there is a glut, and he replies: "I might say that, if I hadn't tried to buy capacity lately."
Thakur says he might ask a long-haul carrier for an OC-48 between, say, Denver and Dallas. "They say, 'Okay, we have capacity, and here's the price,'" Thakur says. "But then, when they launch the order, they may not have the capacity. There may be local loop issues; we may have to build to get to them. You have to worry about whether they're in the same POP that you're in...They may have the wrong address on the building."
The sharpest pain, then, comes in the local loop. Many observers, Thakur among them, say there is a scarcity of capacity in the local loop. "That's because the ILECs still control 80 percent of the local market, and they have been slow to deploy new technologies. Why? Because it's their nature. Unless they have competition or a regulation forcing them to do something, why do it?"
David Schaeffer points out that only a small percentage of buildings--about 5 percent, he says--are connected directly to fiber. Schaeffer is chairman and CEO of Cogent Communications, a Washington, D.C.-based data carrier.
"It is believed that the number of buildings that can be economically served by fiber is probably three times that number," Schaeffer says. "There are a number of economic and regulatory forces that disinsent the deployment of fiber deeper into the metro, and therefore into a larger number of commercial buildings (or data centers and carrier hotels.) Second issue is, many fiber metro providers have yet to adopt a strategy of rationalizing that investment and are not yet willing to sell dark fiber to defray their costs. They'll carry it rather than turn it into cash."
As an example of an economic disincentive to deployment of new fiber in the local loop, Schaeffer cites a carrier's need to sign customers in a high percentage of the buildings it passes with a local build-out.
"Any new entrant only has a fraction of 8 percent of the competitive market, but the incumbent has as 92 percent market share," Schaeffer says. "It's hard to build fiber and then serve one of 25 buildings passed."
The incumbent carrier, by contrast, has an incentive to build out more fiber in the metro core and connect more buildings, Schaeffer believes, but a regulatory disincentive to do so. If the carrier connects a building, it also may have to resell access to it as part of an unbundled loop.
Thakur points out that Time Warner Telecom builds its own metro networks whenever possible. Time Warner Telecom serves 39 metropolitan markets, and plans to serve five more by the end of the year. "Eighty percent of our traffic is 100 percent on our network," he says.
Thakur points out that the carrier also penetrates the metropolitan market via purchases of fiber already laid by someone else. Late last year, Time Warner Telecom acquired the assets (and re-hired some of the people) deployed by the failed GST Telecommunications Inc., gaining local legs in California, the Pacific Northwest and parts of the southwest. With ICG Communications Inc. also in bankruptcy (still operating) and other CLEC bankruptcies widely believed to be on the way, there may be more opportunities to pick up local capacity on the cheap.
"Our philosophy has been that we expand by acquisition or organically (by building), whichever is the best fit," Thakur says.
Gerry Kennedy, vice president-global business development and strategy at Bechtel Telecommunications, says there is demand to be met in the metropolitan networks. His company, a subsidiary of the construction giant Bechtel Corp., is working in 85 cities worldwide, mainly in the United States and Europe, building every kind of local network. But his marching orders have changed slightly in the past year. "In January, the rules were build it as fast as you can, and they will come, and we'll carry your investment out to some form of capitalization such as an IPO and then figure out how to make money," Kennedy says. By year end, it was build the infrastructure, make money, prove it as a business."
Kennedy says he sees the glut/no glut debate more in terms of bottlenecks; like Adventis' Cavallari, he believes there always will be bottlenecks somewhere--in the metro networks today, somewhere else tomorrow.
"We also think that satisfaction of bottlenecks in one part of the market creates demand in other places, so that when we increase capacity in one portion of the net it enables demand to reach other portions of the network," Kennedy says.
In other words, when more fiber capacity comes on line in the metro core, demand will push into the long haul networks, making it feasible to light more long haul fiber.
Maybe then the questions will stop. But don't bet on it.
|
