Is Our Internet Future in Danger?
The digital Disneyland of the future -- where we freely work and play online -- may be at risk. Why? Because, some argue, broadband carriers can't support it. The Internet's "free ride" culture has led to more people downloading gigabytes of data at practically no cost. Even if broadband infrastructure's capacity doubled or tripled, there's no avoiding the equivalent of an abrupt work stoppage.
There are signs of the free ride being nearly over. In the U.K., a million users are about to bump into "soft caps" for usage that their carriers imposed, according to consumer research group uSwitch. In the U.S., some carriers have also started imposing caps that customers have found out about only when they exceeded them in their inaccurately labeled "unlimited" plans. (These limits were hidden in the "unlimited" contracts' fine print.) Comcast, for example, now has a national cap of 256GB per month. And a few are experimenting with tiered pricing, where the more you use, the more you pay -- just like you do for electrical, gas, and water.
Where is all the bandwidth going? Downloading Iron Man in HD, tapping cloud-computing services for business purposes, deploying collaboration technologies that connect disparate workforces, downloading and sharing music and other entertainment files, playing online games, and using remote-work tools like VPNs, VoIP, and app streaming.
"In the space of two, three years, [bandwidth] demand has been doubling and tripling," says Michael Voellinger, senior vice president at Telwares, which helps clients manage telecommunications spending. "And there's only a certain amount of capacity. It's a shared resource."
To be fair, some analysts say there is no bandwidth crisis looming, that there is plenty of capacity available. Derek Turner, research director of the anti-"big media" advocacy group Free Press, says if there were a looming shortage, most carriers would have already imposed tiered pricing or explicit caps; the fact that only a few do points to weaknesses in their specific infrastructure, not to a general shortage of capacity.
But assuming a looming bandwidth shortage -- whether widespread or local to certain areas -- analysts agree that two things must change. First, bandwidth pricing needs to be like any other precious utility, such as water or electricity, which means charging heavy users more to both discourage wasteful usage and bring in money to support such usage. "Broadband is becoming a utility" with a baseline usage cost and overage charges, says Gartner analyst Elroy Jopling. "You wouldn't leave the faucet on and let it run." Second, service and content providers need to rework their networks, protocols, and content to use less bandwidth.
Short of the federal government stepping in -- an unlikely event given historic U.S. aversion to regulation -- the telecom industry will likely adopt its own approach, focusing on increasing revenues from its customers or limiting their use through pricing penalties.
Video usage could bring us to the breaking point
The main culprit today for high-bandwidth usage is video, both user-generated and commercially available. Carriers underestimated the growth in video usage, says Forrester Research analyst Lisa Pierce, who cites an AT&T claim that video counts for 40 percent of the Internet traffic it carries -- up from almost none three years ago. An AT&T spokesperson says that 5 percent of its subscribers account for half its broadband traffic usage, and cites those heavy users as why it's exploring new pricing models.
It's easy to miss how much media services eat up bandwidth. So Gartner analyst Jopling has come up with a rough consumption guide for what his provider's 96GB monthly usage cap supports: 32 hours of interactive gaming, or 48,000 photos, or 900,000 e-mails, or 24 HD movie downloads. Downloading a single high-definition movie eats up 4GB to 5GB, which also happens to be the total amount of data that the average broadband subscriber downloads in a month.
As video is increasingly downloaded over the Internet, carriers could see their customers' bandwidth usage quadruple or more. Video providers are egging customers on: Amazon.com, Blockbuster Video, and Netflix now all offer various movie-download services, including options that stream videos straight to TiVo digital video recorders. To customers, these are simply convenient delivery options; the fact that they consume so much bandwidth is rarely considered.
And today's 10- to 22-year-olds will push the envelope even further, notes Gartner analyst Amanda Sabia. In several focus groups she has run, this cohort says it expects to use video even more than today's typical user but expects to pay no more for the bandwidth. Forrester's Pierce agrees the bandwidth-consumption problem is only going to get worse as next-generation students make their way into companies. They've been weaned on the idea that video, music, and gaming are essentially free and unlimited, and thus act accordingly. "Students live in very wired areas," Pierce says, "which is great for education, but it's not the real world."
The real issue is not video, but the ever-increasing use of the Internet for content and services that take more and more bits each generation, notes Pierce. "What is 'high usage' will be a moving target for the rest of our lives," she adds. For example, Facebook traffic represents about 10 percent of the total Internet usage, she notes. Then there's the plethora of YouTube videos, online gaming, multimedia-rich sites like Disney, and sites like ESPN that automatically launch videos when people visit them, all consuming massive chunks of bandwidth capacity.
"I am concerned that we're seeing a lot of stuff from Web sites being incredibly, overly rich," says Jack Wilson, enterprise architect at Amerisure Insurance, which has a large remote workforce dependent on residential broadband service to do their jobs. "It can even be a problem inside the network. Half a dozen people streaming audio could cause a big bump in our pipe even to one of our remote locations, even to a T1 line."
Contributing to the problem is how content is distributed. For example, peer-to-peer traffic -- whether to share videos, game-playing, or music -- is very inefficient compared to streaming content from a single provider, notes Pierce. The reason is that the network architectures weren't designed for peer-to-peer, which assumes as much bandwidth in each direction, while broadband networks give most of their bandwidth to downloading. That means the uploading channel gets clogged much faster, and the downloading channel may sit empty waiting for the upload to complete.
That's why Japan's telecom ministry is investigating a rearchitecting of its broadband network to work better with peer-to-peer traffic -- emerging peer-to-peer protocols may actually make that sort of traffic more efficient than traditional traffic on such a rearchitected network. But such a rearchitecture effort will be costly and take many years, so even if it is a better model, it won't be in place any time soon, Pierce says.
Companies are also adding to the capacity problem, although not on the scale of video. Many allow more and more employees to work from home, whether periodically or full-time. Remote tools such as VoIP, VPNs, and most notably, videoconferencing, consume bandwidth at little or no cost to the company. Such services don't consume nearly as much bandwidth as video, and the current broadband infrastructure could handle them, notes Gartner analyst Jopling. But video and future bandwidth hogs could get in the way of such services' availability. "It's unrealistic to have all entertainment be delivered digitally [over the Internet]," he says. "You would need fiber to the home [everywhere] to be able to have unlimited capacity."
The pressures against increasing capacity
It's no secret that Americans' appetite for broadband has been enormous ever since cable and DSL service appeared on the menu only a few years ago. Today, more than half of adult Americans have broadband at home, with nearly a third subscribing to a premium broadband service that gives them a faster Internet connection, according to a recent survey by Pew Internet and American Life Project.
The top 20 cable and telephone providers -- such as Comcast, Cox, and Time Warner for cable, as well as AT&T, Qwest, and Verizon for telecom -- boast nearly 65 million subscribers, representing about 94 percent of the U.S. market, Leichtman Research Group reports.
With that level of penetration, the carriers are now moving to open up the pipelines. Verizon, for instance, is investing $23 billion to roll out by 2010 in many parts of the country its FiOS fiber-optic network, which provides Internet, television, and telephone service at near-100Mbps download speeds. Meanwhile, AT&T is spending $4.6 billion to upgrade its network with fiber optics.
Broadband cable providers are also trying to open the pipe. A technology called wideband, whereby cable operators bond several channels together to increase Internet-access speeds, is gaining momentum. This technique can provide speeds of 150Mbps.
But such large investments have been slow in coming, mainly because Wall Street dislikes them. Every dollar on capital improvements reduces carriers' profits, and investors tend to punish capital investments by reducing carriers' stock prices, notes the Free Press's Turner. Because most broadband providers have little or no competition, he says, the Wall Street pressure usually prevails.
Here come usage caps, overage charges, and metered Internet
Despite the distaste for capital investment, carriers have begun investing in it. A big reason is that with the high availability of broadband in urban and suburban areas, there aren't enough profitable new customers left to reach.
Carriers can't raise prices for their current service levels -- current prices are already too high to attract the broadband holdouts. But the carriers also can't lower their prices to attract those customers, since that would reduce their income as existing customers trade down to lower-priced plans, says Gartner's Jopling. "For the foreseeable future, there is a limit on how pricing may increase or decrease. We will only get much lower prices if every home has fiber to it," because that would make capacity essentially unlimited.
That leaves two options: offering premium-priced high-capacity services (a form of tiered pricing) and charging based on usage. Carriers are experimenting with both to bring in more dollars.
Usage caps are the stealth form of usage-based pricing, though several lawsuits have forced U.S. carriers to stop being secretive about them. For example, Comcast agreed this summer to pay Florida $150,000 in a settlement over Comcast's policy of prohibiting excessive use of bandwidth without informing customers of limits. Some customers had their service cut off. In October, Comcast started its new policy, which limits the amount of data a subscriber can send and receive every month at 250GB. Violators will have their service suspended for a year.
Now, Time Warner Cable is testing metered-usage pricing. New broadband subscribers in Beaumont, Texas, are the test case. Pricing starts at $29.95 a month for a speed of 768Kbps and a 5GB usage cap, and it goes up to $54.90 for 15Mbps speed and a 40GB cap. It costs $1 for every gigabyte of usage over the cap. Subscribers can check their usage on Time Warner Cable's Web page. AT&T is testing a similar plan in Reno; the basic $15 768Kbps plan has a 20GB-per-month usage cap, while the as-yet-unpriced 10Mbps plan has a 150GB-per-month usage cap. Users pay $1 per extra gigabyte used, and can check their usage at AT&T's Web site.
But metered billing has quickly come under fire. AOL tried metered billing in the early days of the Internet, and consumers ultimately rejected it; AOL introduced an unlimited usage plan in 1996.
As Time Warner Cable turns back the clock, Free Press's Turner complains, "Its overage fees with rates are completely divorced from what it actually costs Time Warner to provide that data. I don't think you'll see a shift to caps and overage fees -- consumers really don't like surprise bills." Agrees Telwares's Voellinger: "A cap isn't the answer to this. You don't want to limit what people can do."
Broadband and service providers need to be more efficient
It's easy to blame consumers for overusing bandwidth, but it's unfair to single them out.
Consumers do in fact pay for the bandwidth they use, at least partially, notes Free Press's Turner. Content providers pay their Web hosts (or dedicated Internet providers if they host their own Web servers) based on the traffic they use, so customers pay them based on usage. Thus, a video provider pays more to serve up its offerings than a simple Web site does. And this usage-based money moves throughout the broadband ecosystem, with each Web host paying its dedicated Internet provider based on usage, each dedicated Internet provider paying the backbone providers based on usage, and each backbone provider paying the intermediate and last-mile providers based on usage. Even if a content provider like Facebook doesn't charge its customers directly, it is paying for the traffic they use through the fees it pays directly and indirectly to all the infrastructure providers between it and those customers.
However, such indirect payments to the carriers don't foot enough of the bill, says Forrester's Pierce. The providers pay for the bandwidth of what they put into the Internet, but because content and services are used repeatedly, that upfront loading cost represents just a fraction of the downstream consuming cost. For example, if Netflix or Apple uploads a video, it pays for the traffic for that single upload, not for the multiple downloads by users, she notes. That's why carriers are increasingly focused on charging users for their consumption.
Pierce suggests that content providers should pay more upfront, building the cost in to their pricing. She notes that's exactly how Amazon.com's Kindle e-book reader service works: Amazon.com pays Sprint a fee for every e-book sold, since Sprint delivers the content to Kindles over its 3G cellular network. Such a deal was fairly straightforward in the Kindle case, she acknowledges, because only Sprint's network is involved. But Pierce suggests that the same approach could be adapted to the existing system in which content providers indirectly pay carriers for traffic through their Web hosts and dedicated Internet providers. This approach would also have the advantage of allocating the costs to those who actually consume the content.
Regardless of how the usage costs are apportioned, the broadband providers could do much more to better use the capacity they already have, Telware's Voellinger says: "What I see is a need for better optimization of the provider network and broader delivery of broadband capacity. There's room in the bandwidth right now to do a lot more." Forrester's Pierce agrees, noting the carriers are very "uneven" in their network-management capabilities. Because the Internet is a shared network, a carrier that underinvests in capacity or is inefficient in its network management becomes a bottleneck for all the Internet traffic that passes through it. And Internet traffic passes through dozens if not hundreds of networks on its way to the user, so the chances of hitting a bottlenecked segment is fairly high. "To get the best experience, you need to look at the weakest link," she says.
"IP traffic is notoriously unpredictable versus ATM and frame relay," says Forrester's Pierce. So carriers reserve 50 percent of the capacity for peak burst traffic over IP, she says, compared with 30 percent for frame relay. The use of network-management technology would help broadband providers reduce the amount of capacity left "dark" for peak usage, Pierce adds. She believes tiered pricing would make Internet usage more predictable, also helping providers hedge their capacity less and thus free up existing capacity.