An Express Lane?

As the Internet is presently constructed, a nine-year-old child playing an interactive computer game online could get the same priority and quality of service accorded a pair of neurosurgeons attempting to conduct a time-critical medical consultation. And Net surfers could create traffic jams, degrading service for manufacturers sending iterations of new-product designs across continents. Now a project called Internet 2, directed by the University Corp. for Advanced Internet Development (UCAID), may give business and academia their own express lane on the Information Superhighway -- and more. Although many companies today are captivated by the Internets time-saving, cost-saving, and information-sharing capabilities, they also realize they need better performance from the network. Greater speed, greater reliability, and enhanced security all are requirements before many businesses are willing to trust a public network. Moreover, there are many increasingly sophisticated, time-sensitive, and bandwidth-intensive applications that could benefit from an improved network, including those for automotive, the pharmaceutical industry, and medicine. Enter Internet 2. A collaborative effort of 130 research universities that make up UCAID, its goal is to create the broadband application, engineering, and network-management tools necessary to enable advanced research and education. Although its motivation clearly is to provide higher-education institutions with advanced networking capabilities, even in its early stages Internet 2 illustrates how collaboration between industry and academia can ultimately benefit both. Internet 2 was born in October 1996 as a grassroots reaction to increasing traffic and congestion on the public Internet. Shortly after the birth of the World Wide Web in 1993, the National Science Foundation (NSF) began to work on the next iteration of Internet technology. Working jointly with network provider MCI Telecommunications Corp., the NSF launched a new backbone network, called the very-high-performance Backbone Network Service (vBNS), to provide a better place for universities to conduct research. However attractive, the vBNS had one problem: It could be used only by NSF-approved institutions. That left scores of other leading universities to contend with the crowded public Internet. "When NSFNet went commercial, the research community was left at a disadvantage," explains Charles Lee, vBNS program manager at MCI. "We opened a program called Connections to allow NSF-sanctioned institutions onto vBNS, and Internet 2 was formed later to give other institutions a high-performance backbone." Although 53 UCAID member universities have received grants from the NSF to use the vBNS, the remaining members will rely on the new proposed backbone, called Abilene, that will be provided by Qwest Communications International Inc., Cisco Systems Inc., and Canadas Nortel Telecom Ltd. The first improvement Internet 2 will test is greater bandwidth capacity. Extra bandwidth will make it easier to download large graphics, sound, and video files. Even so, bandwidth alone wont satisfy the demands of the most advanced organizations. Corporations wont be able to trust the public Internet with mission-critical applications until the network is able to allow for prioritization of traffic. Accordingly, Internet 2 will implement varying qualities and classes of service. On the vBNS backbone, MCI has tested and recently begun deploying a quality-of-service feature that facilitates bandwidth reservations. Users effectively signal how much bandwidth theyre going to need, and its reserved on the network for a scheduled time. Similarly, using the Abilene backbone, Internet 2 researchers hope to go one step further, implementing different classes of service. This differentiation would take place inside a network switching device called a router, through a process called buffer management. Different types of data would be assigned varying class labels. These would dictate the urgency of information, with the information packets being processed accordingly. "Low-delay packets like those from a medical tele-consultation will be moved to the front of the queue at the routers, while packets from a data-mining session, which can tolerate some delay, will move to the end of the queue," explains Stephen Wolff, Ciscos executive director of advanced Internet initiatives. While they wait for researchers to solve the Internets inefficiencies, many companies have been building corporate intranets and firewalls, effectively creating their own internal Internets. But Douglas E. Van Houweling, president and CEO of UCAID, believes that intranets are duplicative. "Setting up an intranet is not that different from installing a private telephone network," he says. "Companies trust the public telephone system now, so theres no need to do that. Eventually, Internet 2 will eliminate the need for intranets because companies will be able to trust the public network, and that will simplify networked computing." The game plan for developing applications for Internet 2, Van Houweling says, is to test new applications in academia before translating them for use in the commercial arena. Internet 2 is just beginning to select its first partners, but on the vBNS backbone, which has been operational since 1995, several companies in partnership with the National Center for Supercomputing Applications (NCSA) at the University of Illinois at Urbana-Champaign are already testing applications. One of the most popular applications is called tele-immersion, in which geographically dispersed users share an immersive experience in a virtual, 3-D environment. This technology, known as the Cave, allows users to manipulate molecules, alter the shape of an airplane wing, or move parts around on an automobile. In one of the earliest tests of the technology, Caterpillar Inc. is using the Cave to design new earth-moving machines. "Normally, a manufacturer like Caterpillar would have to bring all of its engineers, partners, and suppliers on location to view a new design," says Larry Smarr, director of NCSA. "But using the Cave, Caterpillar feeds its blueprints into a computer and the engineers can go through a whole series of what-ifs in a virtual environment." In another example, an auto manufacturer (whose name was not disclosed) is using linked Caves to connect its design teams around the country. "If one person moves the gas tank, the entire model is adjusted in real time to reflect that change. Its a dynamic engineering environment that collapses the development time from months to days," Lee adds. Although existing applications for high-performance networks are very much experimental, researchers are confident that software applications developers will make good use of the functionality theyre providing. "If you create bandwidth and advanced functionality, someone will figure out what to do with it," concludes Wolff. Qwest President and CEO Joseph P. Nacchio agrees. "Im concerned with providing the bandwidth for Internet 2," he says. "I dont know what new applications will be built to use this bandwidth, but I know there are many smart people out there who will create those applications." Fuel Tech Inc., a Naperville, Ill., company that develops pollution-control technology, has been working since 1994 on one of the most advanced applications for high-speed networks. The companys scientists use computational methods to build virtual boilers that predict what levels of nitrogen oxide reduction can be achieved under various circumstances. But as they add more and more data to the computations, the scientists need more supercomputing power to handle the complexity and more high-performance networking technology to connect them to the supercomputing centers from remote locations. In response to these demands, Fuel Tech partnered with Argonne National Laboratory in late 1994 to develop next-generation virtual-boiler technology. Today Fuel Tech has tested a system that uses two Caves, a supercomputer, and the vBNS backbone that will allow its scientists to collaborate in building increasingly sophisticated predictive models. In the future, says William Michels, Fuel Techs manager of advanced computing systems, the company will be able to automate the entire process of implementing its pollution-control system, "so that well be able to do a project in a day or two rather than six to eight weeks." Cost, however, may be a significant inhibitor. "The technology weve developed with Argonne is actually ready to be used commercially, but its very expensive and our market really isnt large enough to justify the purchase," Michels notes. "Instead, were looking closely at business opportunities to share the costs of owning the technology." Despite concern among some companies that the next Internet will be for researchers only, Internet 2 and the concurrent developments at the NSF seem to suggest otherwise. In fact, industry involvement is a necessity for advanced networking technology, say those involved. "A network that just ties us [higher-education institutions] together wouldnt do the job," says Van Houweling. UCAID is evaluating concepts for Internet 2 applications development and selecting a group of partners to participate in the networks next phase. Van Houweling says hes looking for companies with long-term visions for ways high-performance networking capabilities could help them. "What were doing wont help potential partners in the next six months," he cautions. The flip side, though, for those who are patient, sounds alluring: "Internet 2 equips its corporate partners to be technical leaders in the next two to five years." Nikki Goth Itoi, supplements editor of the Red Herring magazine, can be reached at [email protected].