Daily, billions of individuals across the world wake up, boot up, and sign on to the Internet. What started as a simple exchange of test data among computers in a UCLA lab on September 2, 1969, has morphed into one of the world’s most significant technical achievements.
Despite this unprecedented success, research continues on ways to improve the Internet’s performance and capabilities. One area of focus is on technologies that could someday replace the TCP/IP protocol stack, something that has triggered a great deal of discussion and debate among the network’s most prominent developers, including Vint Cerf.
Whether or not TCP/IP is headed to the scrapheap is unclear, but according to Darleen Fischer, program director for networking technology and systems at the National Science Foundation (NSF) , which awards about $40 million per year toward Internet research, work on merely incremental Internet improvements invites stagnation.
New thinking is needed. TCP/IP, which divided communication tasks into distinct sectors, such as link, network, and application layers, is being rescrutinized. Its approach made sense 40 years ago, but the lines dividing applications, network protocols, and communications links have blurred through the years. Recently, the NSF has challenged researchers to focus on a clean-slate approach and develop networking technologies not bound by TCP/IP’s limitations.
One such attempt to develop a next-generation protocol stack is the Global Environment for Network Innovations (GENI). This project has lured computer scientists who are trying to develop a protocol stack that is not as rigid as TCP/IP and that supports cross-layer communications, so information can more easily flow from a network connection to an application. The group also plans to design a new control-and-measurement framework so ISPs and enterprises can manage their Web protocol stacks more effectively.
The NSF chose BBN Technologies (recently merged with Raytheon Co.), to oversee the project. Many of the GENI researchers also hail from organizations that belong to the Internet2 project, which has contributed network capacity to GENI.
The first prototype GENI core network nodes have been installed within two Internet2 backbone sites and have begun to be tested. The nodes were created by the ProtoGENI team at the University of Utah, led by Rob Ricci, and the Internet Scale Overlay Hosting team at Washington University in St. Louis, led by Jon Turner and aided by Chris Tracy of the Mid-Atlantic Crossroads (MAX) consortium, a group established by Georgetown University, George Washington University, the University of Maryland, and Virginia Tech to provide research and production testing services in Maryland, Virginia, and Washington, D.C.
With the network nodes in place, various experimental networking projects can run tests based on the GENI work. One of these projects is PlanetLab, a consortium for technology development that is managed by Princeton University, the University of California at Berkeley, and the University of Washington, with support from the NSF, Intel Corp. (Nasdaq: INTC), and the U.S. Defense Advanced Research Projects Agency (DARPA).
One of PlanetLab's main purposes is to serve as a test bed for new networking protocols and GENI network designs. More than 600 projects are underway, and participants are examining a variety of what PlanetLabs staffers call “planetary-scale” services, including massively scalable file sharing and network-embedded storage; content distribution networks; routing and multicast overlays; QoS overlays; scalable object location; scalable event propagation; anomaly detection mechanisms; and network measurement tools.
Even if this work never results in wholesale replacement of ISP networks, it's sure to boost the development of at least a few future improvements to the Internet.
— Paul Korzeniowski is a freelance writer who has been dissecting technology and business issues for two decades.