market intelligence

In search of the big pipe?

For Internet service providers, timing is everything when it comes to bandwidth acquisition.

BY ROBERT PEASE


Depending on which telecommunications analyst you ask, data traffic is growing between 30% and 80% per year and will dominate worldwide traffic sometime in the next few years. It is now an accepted fact that data, spurred by a world that's embracing the tremendous versatility of the Internet, will surpass voice traffic sometime in the next decade. Many experts contend it already has.

Internet service providers (ISPs), in the form of legacy telecommunications companies, startup opportunists, and everything in between, are finding themselves facing multiple challenges in fulfilling demand for their services. One would expect that obtaining sufficient bandwidth to run an Internet service would be the primary concern. But availability of bandwidth alone, for most ISPs, is the least of their worries. Finding inexpensive bandwidth at just the right time in just the right place-now that's another issue.

In looking at the bandwidth demands of ISPs, it is dangerous to lump all providers together. A differentiation must be made between local, regional, and national ISPs. Small, local ISPs seek only to buy or lease their Internet connectivity from a larger regional or national counterpart and aren't necessarily building out their own backbone network. On the other hand, regional and national ISPs are continuously shopping for connectivity that will enable them to build proprietary backbones.

There are currently about 30 U.S. companies acting as national Internet backbone providers. Each has connectivity to the four Network Access Points (NAPs) located in San Francisco, Chicago, Washington, DC, and Pennsauken, NJ. The interconnection of the NAPs, which are basically very large switches owned by various large telecommunications companies, forms the primary Internet backbone. The same types of exchange points are located globally, though in some countries they are tightly controlled and regulated by the government.

The task for the local ISP is to get connected to the nearest NAP by "shopping and comparing" prices based on its location. Sometimes a bandwidth broker will be employed. Bandwidth brokers specialize in selling excess capacity from facilities-based carriers to ISPs.

"For instance, a local Texas ISP that does business in two or three markets in Texas is typically not going to buy fiber to all the major Internet exchange point to connect with all the networks directly," says David Williams, vice president of strategic marketing at GST Telecommunications Inc. in Vancouver, WA. "Rather, they'll just buy a DS-3 from a larger ISP and focus on their local connectivity and other services they provide. As they get larger, they may begin to move to direct connections to major interconnect points."

Williams, who was recently elected chairman of the board of the ISP Business Forum, a national organization representing progressive ISPs, says the concern for the local ISP is not bandwidth availability. Rather, it's in finding the capacity and interconnection at a price low enough to get a return on its investment. That's where location enters the equation. In areas of fierce competition, it's more of a buyer's market and bandwidth can be obtained at a lower price. But in places where only one or two providers offer an "only game in town" bandwidth package, ISPs can find themselves paying top dollar to provide services-and that high price is passed on to consumers.

Ovum, an independent research and consulting company, refers to these different competitive environments as "fat routes" and "thin routes." A fat route is one that has a large level of demand and, correspondingly, a large amount of investment in the supply of capacity through new infrastructure and lots of competition. A thin route, conversely, has less investment and less competition, resulting in less incentive for existing players to discount capacity. It's also a cyclical pattern in most cases, where thin routes, over time, become the fat routes.

"The best example, on an international scale, is the transatlantic route," says Barry Flanigan, senior analyst at Ovum. "A few years ago, the transatlantic route was running out of capacity for carriers and ISPs. One of the reasons was that the guys putting in the submarine cables traditionally got the forecasting wrong and underestimated the extent to which Internet traffic would grow. Then, of course, we saw phenomenal growth rates in Internet traffic, leading to a corresponding shortage of capacity in the transatlantic route.

"This led to a huge investment in supply of more capacity, and two more cables, Gemini-1 and Atlantic Crossing-1, went in to meet the need. Since then, another round of investment has resulted in a lot of new transatlantic cables coming on line in the next year or so. You've got TAT-14, Atlantic-1, Level 3, and Atlantic Crossing-2, to name a few. So you have a situation that has actually flipped over from a capacity shortage to this potentially huge overhang of capacity and, correspondingly, you've seen a tremendous pressure on prices across the transatlantic route, so bandwidth prices will be tumbling."

Purchasing decisions for ISPs are also much more complex than in the past. In most cases, competition and availability create a greater variety of options for purchasing capacity. In the past, when bandwidth was fairly limited, telecommunications carriers owned the fiber-optic backbones, using whatever capacity they needed and selling the excess to buyers at premium prices through 15- to 20-year indefeasible rights-of-usage (IRU) agreements. Today, with technological advancements like dense wavelength-division multiplexing, capacity is plentiful and can be obtained in many cases on a short-term basis, as little as one year for some systems.

"DWDM is having a huge effect on bandwidth availability," says Flanigan. "The supply-driven model, using DWDM, is perhaps the most important recent development. It has provided the key to unlock the ability to access huge amounts of fiber capacity. It also provides more control over the way a provider upgrades capacity, allowing upgrades in a much more scalable fashion. Once you put the fiber in, even though faced with the unpredictable growth of data and Internet traffic, you can still grow your network as the demand grows. In terrestrial networks, capacity can be upgraded almost indefinitely."

Making choices

So considering fiber is fiber and bandwidth is bandwidth, and both are in ample supply, for ISPs it comes down to where they locate their services, how cheaply they can buy bandwidth, and what routes they prefer to use. There really aren't a lot of value-added services when you talk about raw bandwidth or raw fiber connectivity, says GST's Williams. So what separates one ISP from another and, with decreasing bandwidth costs, will more players continue to jump into the ISP market?

The question of which ISP offers more than the next may be less a question of services offered and more a question of the technology behind the services themselves. "If you start moving into ATM [Asynchronous Transfer Mode] switch connectivity, there may be some room for differentiation," says Williams. "Still, once you move up to the Internet connectivity layers, there is a lot more room for differentiation. But at the fiber level itself, it's really location and price."

Williams says that although there is still a lot of room for niche players and differentiation of services in the Internet industry, a lot of consolidation has taken place over the years, and that trend will likely continue.

"We're already seeing large mergers, such as MCI and WorldCom, that pool an enormous amount of backbone Internet traffic through one company's facilities," says Williams. "Now, with the merger of Sprint and MCI WorldCom, that company will control a vast majority of the Internet backbone traffic-not necessarily the underlying fiber, but the traffic that traverses what we know as the Internet. They're providing the network, along with the connectivity between all the other networks, over their own infrastructure. About 70% of all Internet traffic in the United States will likely be controlled by the combination of MCI WorldCom and Sprint. That means about 70% of all ISPs will get their Internet connectivity from the combined Sprint/ MCI giant, which is causing considerable concern in much of the industry."

But despite the tremendous consolidation efforts taking place, the number of ISPs continues to increase each year. For each one bought, says Williams, there are probably 10 startups, due to the potential for differentiation and the number of existing under-served markets. For example, smaller markets tend to be overlooked by large carriers that focus on the top 30 to 50 markets. That results in ISPs popping up in the smaller markets to focus on particular consumer segments and vertical markets, although they still will likely buy services from the larger backbone players.

Perhaps another question concerning available national ISP connectivity might be how rapidly legacy carriers are converting traditional voice networks to handle a combination of voice and data. According to Williams, it really shouldn't pose a huge problem for most traditional voice carriers.

"For instance, by installing ATM equipment, they can run Internet data and voice easily over the same fiber network," says Williams. "Several traditional carriers are migrating from multiple legacy networks to a single converged network. A lot of time, SONET [Synchronous Optical Network] is still the underlying protocol. They can put ATM on top of that and then IP [Internet protocol] and voice services over the ATM. But technology is maturing to allow IP directly over SONET or IP directly over DWDM. There's a big move to DWDM, but migration to new technologies depends on the new technology being proven and reliable for our customers."

But even with bandwidth price dropping and availability increasing, the most important driving force is the rapidly increasing need for bandwidth. The voracious appetite for bandwidth is spreading across the globe, and the ISP that has enough bandwidth today may not have enough tomorrow. It's really a race of sorts to see if bandwidth availability can stay out in front of the insatiable consumption of bandwidth-intensive applications that are emerging today and will emerge in the near future.

According to Ovum's research, total Internet connections alone will grow from 56 million in 1998 to 224 million in 2005. If that factor is not enough to drive bandwidth demand skyward, Ovum also forecasts the growth in bandwidth-intensive applications to further drive the ISPs' continuing demand for additional bandwidth. New bandwidth-hungry applications include real-time video feeds via the Internet, large file attachments to everyday e-mails, dynamic Web pages with video and graphics as opposed to static Web pages, and interactive multimedia and desktop multimedia networking. The future, according to Ovum, may hold a lot of surprises regarding ISPs and the Internet.

"In 10 years' time, the telecommunications environment will be characterized by low-cost, high-capacity international transmission facilities encircling the globe," says Stephen Young, an Ovum analyst, in a recent "thought paper" addressing the bandwidth explosion. "Surges of traffic will be readily accommodated because networks will have enormous capacity and multiple routes available. Networks will be built on IP-based infrastructure rather than traditional circuit-switched infrastructure. By using DWDM, carriers will build networks that offer unlimited bandwidth at a fixed price between any two points. Distance and geography will become irrelevant factors in telecommunications costs and pricing."

Who will be the ultimate winner in such a bandwidth-rich world? Perhaps the ISP will benefit to an extent, but the real advantage will be passed to the end user who will have access to cheap, plentiful capacity. In fact, Ovum believes the successful and unsuccessful players still have yet to be determined, but "there will be no returning to the old environment of bandwidth scarcity and artificially high tariff structures."