PDF of this tutorialThe Internet continues to be prone to significant delays and congestion that frustrate all users, even those on broadband connections. Performance is so poor that content providers are able to offer only the most rudimentary applications and services. The average speed of the Net is 80 times too slow to offer full-motion video, and the problems are only getting worse as more people use the Internet. The speed of one's modem—or the first and last mile as it is known—has been a significant factor in the development of more bandwidth-heavy applications. Modem speed places considerable restrictions on a content provider's ability to offer applications and services requiring faster speeds. As broadband penetration rises and consumers have access to faster modem speeds, these speed restrictions will be significantly alleviated.
Speed and reliability are the two most common problems cited by Internet users. The typical Web page contains 15 to 20 kilobits (Kb) of information, so it will not overload the network. In comparison, one second of video contains more than 125 Kb of information, or close to 10 times that of the average Web page. Thus, the speed and reliability problems of the Internet limit the applications that run on it. Median download times on the Net are close to 3.5 seconds, while mean times are approaching 10 seconds. Given the limitations of the Internet to deliver reliable high-speed connections, current Web pages are designed not to exceed 15 to 20 Kb of information. This is to ensure that sites can be downloaded within at least a few seconds. In a recent Dataquest study, they found that streaming media often travels through 20 different routers and experiences 25 percent packet loss. (Packet loss occurs when a portion of the signal does not reach its destination the first time it is sent). The further a packet must travel, the more likely it is to run into delays.
Today's DSL networks offer only a slight improvement because they only affect the first- and last-mile bottlenecks, and not the router and server bottlenecks. DSL and cable modems are great at resolving the first- or last-mile bottlenecks, but they simply push the bottleneck further upstream. Streaming applications put more stress on the network because they demand more bandwidth than the 25 Kb per subscriber available. Local caching is required to guarantee a positive user experience and to differentiate broadband content from others. To eliminate delays and improve the way content is delivered on the Internet, all aspects of the process must be improved.
The other two bottlenecks, router and server congestion, are being addressed by a number of service providers such as Akamai, Digital Island, and Cidera. Moreover, new routing technologies, faster equipment, content-distribution systems, and caching are helping to address these problems.
Despite the bottlenecks, the Internet and the increasing penetration of broadband access are quickening the spread of new applications and of new technologies. In less than one year, Napster attracted more than 38 million users, proving that the market for peer-to-peer and other file-sharing technologies is huge. Since then, hundreds of new Internet start-ups have been seeking to capitalize on this success. Napster was a success because it brought people back to what they liked most about the Net—the ability to connect people with people, even if it was to exchange music without having to pay copyright fees. According to recent studies, when asked whether they would give up e-mail or the telephone, consumers were equally split as to which they would give up. However, when a similar choice was offered between the Web and e-mail, more people said they would give up Web access. This statement is just as true for individuals and large corporations. E-mail is what makes enterprises run.
Andy Odlyzko, in a recent paper titled "Content Is Not King," found that people are willing to pay a lot more for point-to-point communications than for anything else. E-mail is the killer application precisely because it is all about one-to-one communications. While content can be profitable, as is seen by the amount of money earned annually by media companies, content by itself is often not a profitable endeavor. Every newspaper, with the exception of the Wall Street Journal, has failed in their efforts to convert their readership from a free model to a subscription-based model. Inexpensive Web browsing, even if it is at higher speeds, is not what consumers are looking for in the Internet. What they really want is the ability to interact with other people.
The increasing popularity of sharing pictures or videos and the cumbersome nature of e-mail in sharing these photographs will lead to an increasing use of peer-to-peer networks that will be specially designed for this. The growth of local storage and the increasing availability of tools to fill up this storage with video clips and other material will likely make peer-to-peer networks increasingly important. Although it is difficult to determine the path the Internet will take, one can be reasonably sure that the growing storage and communications capacities that exist and are in the process of being developed will be used often in unexpected ways. Napster came out of nowhere, and the next file architecture could also. Odlyzko predicts that as the Internet becomes ubiquitous, point-to-point communications will provide higher value than broadcast communications. As for the question of whether the Internet will continue to become an open network or whether peer-to-peer networking will eliminate the openness of the Internet, Odlyzko argues that effective point-to-point communications will demand easy interconnection and openness—two items that can only be provided if the Internet remains open. He predicts that we will likely end up with a system where commercial users contribute the most, but households paying for point-to-point communications will be close behind, with the transport component of charges for content coming in last.
