That giant sucking sound you hear is the collective gasps of enterprises and service providers that were expecting the massive build-out of the public network to produce bandwidth in quantities as plentiful as the air they breath.
Despite recent predictions of a bandwidth glut due to construction of fiber networks by Enron, Level 3 Communications, Global Crossing, Qwest Communications International, Williams Communications and others, many experts believe a bandwidth surplus leading to cut-rate telecommunications costs will not materialize in the near future.
In fact, some analysts are predicting a shortage.
"In the next five years we don't see any ability of service providers in the U.S. to keep up with the demand," said Mouli Ramani, director or strategic marketing for the optical Internet at Nortel Networks. "I don't see any chance of getting into a glut anywhere in the network over the next five years."
At the request of Nortel executives, Ramani has spent the past year cooking up extensive calculations, which project demand for bandwidth will increase more than twice as fast as carriers can increase capacity. Based largely on Ramani's findings and growing demand for networking gear from service providers, Nortel last week pumped US$260 million into expanding the capacity of its optical equipment manufacturing facilities, the second such expansion for the networking giant in less than three months.
Others seem to be following Nortel's lead.
Optical networking component maker Corning last week made two strategic acquisitions worth more than US$2 billion and launched a joint venture with Samsung Electronics to ramp up its ability to meet the increasing demands of its customers. Earlier this month, Lucent Technologies shelled out nearly US$3 billion for optical laser manufacturer Ortel.
Equipment makers are scaling up production to meet the needs of their service provider customers, which are racing to activate, or light up, dormant strands of fiber-optic cable to meet existing bandwidth demands and accommodate expected surges in data traffic.
All of this activity comes against a backdrop of a widely held perception that the networking industry is headed toward a bandwidth glut. That belief is the result of several articles that appeared in national publications in 1999 suggesting the thousands of miles of fiber put in the ground by Level 3, Qwest, Williams and other so-called green-field carriers would produce an abundance of bandwidth in the next few years.
Citing studies from research firms including Forrester Research and Renaissance Worldwide, the articles said the resulting surplus could push pricing to record lows and threaten the viability of several players. Although Renaissance recently reduced its projected differential between bandwidth supply and demand, based on service providers' ability to activate dormant fiber, the research firm still predicts a bandwidth glut in the long-haul portion of the network.
"Everything we're seeing is just the opposite," said Lynn Hutcheson, a research analyst at Ryan Hankin Kent. "If you're looking at how the amount of data going over this network is growing, it just doesn't seem like there will be a bandwidth glut over the next years."
The unpredictability of the Internet's growth helped make the feast-or-famine debate a contentious issue for the past few years. The glut vs. scarcity pendulum, however, appears to be moving back toward the scarcity side.
"If you believe the Internet is going to be the dominant force driving bandwidth demand, then you realize that no matter how big a pipe you have, you will fill it," said Jian Li, senior director of emerging technology at Qwest.
Experts routinely point to at least three factors influencing future demand for networking bandwidth. The first is the expected increase in the size of the pipes connecting to the Internet. As broadband access technologies such as Digital Subscriber Line (DSL) and cable modems reach mass deployment, they could exert tremendous pressure on public networks.
Tom Nolle, an analyst at research firm CIMI, said the shift toward broadband access has the potential to dramatically magnify the amount of traffic on the public network. If regional Bell companies follow through on plans to deploy DSL services, the network load could exceed the current traffic loads by a factor of more than 10.
In addition to increases in the size of the access links, the number of users is expected to grow as the Internet continues to penetrate popular culture - by as much as 25 percent annually, according to estimates.
A third factor that will increase network traffic is new applications, such as video programming and collaborative work sessions. Nortel's Ramani said these types of applications will increase the amount of time users stay online and the amount of data they exchange. Built into most projections is also the assumption that new applications will emerge that consume even more bandwidth. "Keep in mind that bandwidth and applications are in a chicken-and-egg situation," said Tom Valovic, an analyst at International Data Corp. "I'm a firm believer that if you put bandwidth in, you'll stimulate the market in new and significant ways."
Based on these factors and other online usage trends, Ramani calculates bandwidth demand will grow to 100 to 200 times today's rate in the next four to five years. At the same time, even if next-generation equipment makers were able to light their total arsenal of fiber in the next five years, capacity would only increase by about 70 times existing levels, he said.
It is the failure to appreciate the cost and difficulty associated with the provisioning of raw fiber to carry data that may actually have attributed to the inaccuracy of some bandwidth predictions, according to Qwest's Li. There is a misconception, he said, that raw fiber equals bandwidth. Qwest and others have buried thousands of miles of fiber, as well as empty conduits to house thousands of miles more of the hairlike glass strands. That so-called dark fiber, however, represents potential bandwidth. It takes massive amounts of money and time to turn it into usable capacity.
Ramani said predictions of bandwidth surpluses also fail to acknowledge that the public network is segmented into several regions. "They treat the entire network as if it's one big black box," he said.
Most analysts point to the network's access edge, which is loaded with equipment designed to transport mostly voice traffic in increments of 64 kilobits per second, as the major choke point. Ramani said Nortel and others are pouring massive amounts of optical gear into that region to create a wider on-ramp to the network's core.
Even analysts supporting the theory of a potential bandwidth glut in the public network's core acknowledge the widespread scarcity of bandwidth at the access portion of the network. "If you look at what's there [in the long-haul portion of network] now, and the ability of the providers to saturate it, it's way overbuilt," said John Morency, an analyst at Sage Research. "It's a matter of how quickly you can ramp up access to drive the backbone."
Ramani, however, said the attention now being focused on the network's edge will create future bandwidth scarcity at the Net's core. "Once those bits fill up at the edge of the network," he said, "they have to go someplace."
