NSF focuses on high-speed networking applications

A year after dismantling its flagship NSFnet the National Science Foundation announced that it will focus on the development of high-speed networking applications - a move designed to relieve researchers fuming over congestion on the Internet.

Specifically NSF is entering the second phase of development for its very high-speed Backbone Network Service (vBNS) the next generation of high-speed networking for scientific use. vBNS will increase connectivity from the current five NSF supercomputing centers to as many as 100 connection sites and could yield as many as 1 000 new advanced applications in areas such as virtual reality digital libraries and visualization.

Networking applications such as these sophisticated connections require more bandwidth than that offered by the commercial Internet. NSF is touting vBNS as a "productivity" network that will serve as an on-ramp for emerging technology bound for the Internet. The agency is dodging participation in any applications that could be fully developed in the open market.

"The information highway is being broken into regular and express lanes that support new applications which do not work well on regular Internet " said Mark Luker NSFnet's program director. For access to these high-speed lanes however an application must be more than "just lots and lots of traffic " said Luker who added that the applications now belong to fields such as weather and nuclear physics.

For this phase of vBNS development NSF has funded 13 university proposals for "innovative high-performance connections." Each will receive $350 000 in seed money to develop applications and connectivity 35 additional sites are in the works. Additionally NSF is working with agencies such as the Defense Advanced Research Projects Agency and the Energy Department to establish vBNS links to federal research networks.

Congress has historically hounded the agency to stay out of commercial development. "In congressional hearings and the like there has always been a concern that NSF is crossing the line in terms of what is appropriate in the networking area " said Brett Berlin a consultant with Berlin and Associates. "NSF has become very sensitive - and properly so - as to what the right role of a government agency is in this area. This is an honest attempt to strike the balance in an area where there is no magic formula but an ebb and flow a give and take of effort."

MCI Corp. is helping to develop vBNS under a $50 million cooperative agreement. "This is consistent with NSF's role " said Diana Gowen director of Defense Department sales and marketing at MCI Government Markets. "It is purely research. vBNS connectivity is not there as a replacement of the commercial connectivity that these research entities are required to have."

The addition of traffic to vBNS will eventually lead to further MCI development of advanced commercial applications said Charles Lee senior manager at MCI Government Markets. "We are setting aside bandwidth in a dynamic environment. If we learn how to do that most efficiently that gets migrated into our commercial offerings " he said. "Now all of a sudden we have to start dealing with live traffic in a congested environment."

The kind of bandwidth promised by vBNS - initially 155 megabit/sec with a boost to 622 megabit/sec or OC12 - will be a godsend to researchers said Ahmed Kassem director of academic computing and networking at the University of Illinois Chicago - one of the 13 campuses chosen by NSF.

"By relieving the problem of congestion this opens a new horizon. Connectivity stimulates research. If you have a bottleneck you spend your life trying to get around the bottleneck " he said.