A clean slate for the Internet

Funding the slate

The National Science Foundation has funded Internet research for many years, but most of its projects have been of the incremental improvement variety, and most have not involved proving out new ideas on a large scale, with millions of users, says Deborah Crawford, deputy assistant director for computer and information science and engineering at the NSF.

But now the NSF is gearing up to build a US$300 million to US$400 million clean slate on which researchers can chalk up and test radical new ideas. The Global Environment for Networking Innovation, or GENI, will be a giant test laboratory stretching across the U.S., complete with wired and wireless computers, routers, switches, management software and subnets of wireless, cellular, sensor and radio devices. It will include a fiber-optic backbone and tail circuits to some 200 universities.

When it's complete, sometime after 2010, users will contract for virtual "slices" of the GENI infrastructure, which they'll be able to use to test ideas "at scale," simultaneously and independently, Crawford says. Each researcher will see his slice as his own private test network.

GENI will be built without assuming anything, says Allison Mankin, a co-manager of GENI at the NSF. The hardware and software will have the flexibility to accommodate the trial of just about any networking idea, not just those based on packet switching, TCP/IP, routers and other accoutrements of today's Internet.

Mankin says the kinds of incremental progress that have typically come from earlier NSF projects, while worthwhile, are no longer sufficient. "For example," she says, "people have actually proven that it's impossible to prevent denial-of-service attacks with the current Internet. If you want to build a network without denial of service, you have to start over."

But Mankin acknowledges that it's unlikely the old slate will be wiped clean completely, with the global Internet scrapped for something entirely different. "Forklifts don't exist that are big enough for that," she notes.

GENI will provide a clean slate for design and large-scale testing, but not necessarily a clean slate for deployment. Says Mankin, "You get a wonderful idea out of the [GENI] facility, and then you say, 'Well, now that I understand how it works, I see how I could put it into the existing Internet because I just have to tweak it, and it'll fit in.' "

In parallel with the design and building of GENI, the NSF is funding a number of research efforts under a program called FIND, for Future Internet Network Design. They are the kinds of efforts that may ride on GENI when it's available, but many are under way now.

One FIND project focuses on how to build manageability into routing systems from the beginning, instead of adding it later as is typically the case. Another explores ultrafast optical technologies, based on the belief that electronic switching systems will be unable to keep pace with mushrooming demand. Still another is looking at "location-aware networks," in which the IP address is supplemented by geographic coordinates. And then there's a FIND project that is considering the option of allowing users to select the paths their packets take through the network, letting them make trade-offs among factors such as cost, speed, reliability and security.

Not all the projects focus on technology. One is studying the interactions and economic and social motivations of users, hackers, service providers and regulators, with the goals of improving service, security and user utility.

Mankin says the Ethane initiative is part of a family of FIND projects at Stanford that could lead to "an Internet where denial of service can't happen." But, she adds, it's too early to tell whether the school's ideas represent a "leap."