20 Dec Landweber on networks, open source, and NSF budget cuts
Larry Landweber knows networks. The University of Wisconsin-Madison professor of computer science, a department member since 1967, helped forge some of the early international channels of the Internet. He also has an idea of where networks are heading – and it’s all about finding the right applications to harness increased computing power.
Before conducting an interview with WTN from Washington, D.C., where he works for the National Science Foundation, Landweber spoke at the Digital Government Summit and Accelerate Madison on the future of networks.
Here are some of his visions and questions for the future, on broadband, open source, the NSF and networked devices.
On broadband
The network is boring. That’s the word from a network pioneer: Landweber said most people only see the applications that run atop the network, not the infrastructure itself.
Soon, though, the infrastructure may outgrow the applications written for it.
“We know what people are going to do with T1s, a megabit per second, because that’s what you need if you’re going to be moving images around, or video around,” Landweber said. “It makes it practical to actually deal with graphics, or images, or multi-player games, or videoconferencing. What we don’t know is what people will do if we get up to 100 megabits per second.”
Home users may not realize how much bandwidth is out there. The “last mile” connecting household cable or DSL setups to Internet backbones is much slower than the fibers connecting large institutions. The solution, Landweber said, is bringing fiberoptic connections to the home, or deploying high-speed wireless connections.
But the more fiberoptic lines there are, the more connections must be routed. Routing light isn’t like routing electric signals directly from wire to wire. The router can interpret the light signals, process them internally, and re-create a light beam on another fiber, but that takes time. Or perhaps you could route the light beam itself – a challenging task, Landweber said.
How do you herd light beams? If they go to fixed locations, all it takes is mirrors. But motorized mirrors take precious milliseconds to rotate. The challenge is greater when each fiber is carrying multiple colors of light.
“They want to be able to figure out where to send a beam of light based on its color,” Landweber said. “That doesn’t sound like a revolutionary idea, but it is.”
The common Internet application may not need that speed or bandwidth, but researchers are increasingly pushing terabytes of data (in the recording industry’s favorite measurement system, a terabyte is almost 300,000 songs) across the network. That takes some heavy-duty infrastructure.
That’s why researchers are turning to Internet2, a scientific network separated from the commercial Internet, with high capacity for that sort of data transfer. UW-Madison has a direct link to Chicago.
Regarding Wisconsin’s new statewide data and video network, dubbed BadgerNet 2, Landweber said he didn’t share some of the concerns UW officials have voiced over its ability to serve the university.
“I think UW-Madison is well-taken care of by the fiber link it has to Chicago,” he said. “And UW-Madison is the principal research institution in the state … I’m comfortable that its researchers will be ok.”
On open source
“Whether you like it or not, people get motivated by money and fame,” Landweber said.
Open-source software development may lead to fame, at least in the developer community. But it doesn’t inevitably lead to money – and that’s not the motive of many open-source developers.
Companies do, though, make money off open-source software by serving the business needs that code alone does not. But is it enough?
“Clearly Linux is doing very well, but clearly if you’re a company you don’t buy Linux unless there’s a company behind it,” Landweber said.
“The question I would ask is whether truly new, innovative things are going to come out of open source,” he said, “or whether it’s going to take entrpenurial spirit and somebody willing to put a bunch of money toward coming up with new things.”
On the National Science Foundation
The NSF’s budget was cut by 2 percent this year. The first budget cut for the agency in years puts its budget at $5.5 billion, $105 million below its 2004 budget and $232 less than what it requested.
Speaking as a professor – not an NSF advisor – Landweber outlined some of his concerns about the funding cuts.
“Potentially, if it continues for another year or two with additional cuts it has very serious consequences for the future of the country,” he said. “One of the areas that has been most important in keeping this country vibrant and competitive is results in science and engineering.”
Both faculty and graduate students depend on grants from agencies such as the NSF. In computer sciences, for instance, Landweber said a faculty member might get two months’ salary from an NSF grant but may support two to three graduate students with it. The effect isn’t limited to the computer sciences.
“Biotech is a big, big win for the United States in our leadership. That’s also been true in the computing and information technology fields. My hope is that this is just a one-time cut.”
On the devices ahead
Wireless will be big. That may come as no surprise. Again, though, Landweber said the network – wireless or wired – is really an enabler for applications or devices. And those devices are starting to compete across categories.
“It looks to me like we’re going toward the more general,” Landweber said. “You get the IPod, and suddently IPod has a big disk, and IPod is going to be a database system. it looks like the trend is more toward generalization than specialization. people get an organizer and then suddenly it’s a cellphone. People dont want to buy multiple devices. If you can have an organizer
and a cell phone together, why not?”