Net gurus ponder future of advanced Nets
- 28 June, 1999 12:01
The Internet's deep thinkers gathered in San Jose last week to hail and fret over the fate of a next-generation Net they hope will be for everyone, not just the technologically privileged.
Making that expansive vision a reality depends on the ability of network professionals and others to overcome a number of stubborn obstacles, including network capacity strains, a dearth of quality-of-service (QoS) standards and meddlesome lawmakers, said participants at the Internet Society's ninth annual conference.
"What scares me the most is running out of [network infrastructure] capacity to meet the demand," says Vint Cerf, who just stepped down as chairman of the Internet Society.
Participants in Internet2, a university-driven advanced network, used the global summit to show off their next-generation research projects. These efforts include high-powered video-collaboration applications that would take advantage of real-time global communication but would also put new strains on network resources.
One Internet2 project, NASA's Virtual Collaborative Clinic, combines complex medical imaging with high-speed networks. Using data from MRIs, CAT scans and other imaging devices, doctors around the world can remotely view and manipulate 3D high-resolution images to diagnose patients and demonstrate possible surgical outcomes.
But project backers say a lack of QoS standards and unproven multicast technology are two of the biggest stumbling blocks to getting this type of high-bandwidth application deployed.
"The processing power is there, the bandwidth is there, but there is no quality of service," says Rich Wall, program director for Internet technology at IBM, which participates in Internet2. If you're going to do a remote diagnosis of the brain, you can't drop or delay packets that make up part of the image, he says.
The Virtual Collaborative Clinic uses multicast technology to disperse images around its high-speed network, but researchers say the technology has a long way to go.
"The biggest issue we had to tackle was getting the routers to function" fast enough, says Xander Twambly, a NASA research scientist. "The pass-through of multicast addresses is still too kludgy . . . and if you start getting substantial packet loss, it kills your network."
The International Centre for Advanced Internet Research (ICAIR) is banking on QoS technology for its Video Portal Service, which was also on display at the event. The online video centre, which lets users stream or download training videos and search through them, will deploy the Internet Engineering Task Force's Differentiated Services standard soon to handle QoS. But ICAIR director Joe Mambretti says the jury remains out on which QoS approach will be implemented across the Net.
Experts at the conference insist QoS is more than just a transport problem, too. "You're going to need quality in your desktop systems," says John Gage, chief researcher at Sun. "The quality of service on the Internet is infinitely better than what comes from Microsoft."
Until vendors make net software more manageable, "network managers will be under tremendous pressure to make the network more reliable," Cerf says.
Bandwidth is only a piece of that puzzle, he adds, and contends that server limitations and the low ceiling of today's forwarding and switching gear present more vexing problems. In general, the demand for capacity is so voracious that companies must redesign their networks every year just to keep up, Cerf says.
He also took on the government regulators, saying the Net is too young for lawmakers to start imposing harsh regulations. "I'm concerned about the proclivity of legislators to pass unimplementable laws," he says. For instance, Internet service providers should not be responsible for filtering objectionable content because "it's technologically impossible", Cerf says.
He calls the ongoing Washington, DC, debate over Internet taxation "hopeless", largely because of jurisdictional disagreements across geographic boundaries. However, Cerf says taxation must be part of any greatly expanded Internet because governments will rely on that revenue.
Although conference attendees were at odds about which problems to tackle first, there was consensus about the notion of widespread global expansion.
"It's more important to connect the world than to place a QuickCam on Mars," says Floyd Backes, a 3Com executive. "Otherwise the economic disparities around the world will continue."