Start-up shines light on Net's "dark matter"

One of the latest remodeling ideas comes from start-up XDegrees, which unveiled a technology this week that grafts a little bit of Napster-like file-swapping and a little bit of Akamai Technologies' Net-speeding service onto the old system of Net addresses.

XDegrees' idea has already drawn $8 million from venture capitalists, and it landed company Chief Executive Michael Tanne on a panel at this week's O'Reilly peer-to-peer technology show in San Francisco alongside representatives from industry giants such as Sun Microsystems, Microsoft and IBM. Analysts who have looked at the technology say it's a potential home run--but only if one of the Web's giants buys in.

"It's a compelling idea," said Evan Quinn, an analyst with Hurwitz Group. "If they were to land one of the big four (software companies) as a client, I think it might cause a little revolution...But they need a champion."

A complicated mix of networking and peer-to-peer technologies, XDegrees' technology aims in part to open a whole new universe of computers, files, applications and devices directly to the Web.

Using the company's technology, people could give the photograph on their own desktop computer its own Web address, for example. On the other end of the scale, the idea could provide a boost to software-through-the-Net services such as Microsoft's .Net initiative, analysts say.

The company's plans are among several designed to raise the Net infrastructure system to a level that supports the growing ambitions of new online services. Some of these plans have met opposition from people unwilling to tamper with the guts of the Net, or who think sweeping decisions are being made without enough review.

But many infrastructure innovations are gaining ground fast as companies from Microsoft on down try to push the boundaries of what old networks were designed for.

Pushing the Net's boundaries
XDegrees is part of a growing movement aimed at patching over what some see as limitations in a Net infrastructure conceived long before the Web itself was born. Companies are trying to offer video, multimedia, and even access to remote-controlled software to tens of million of people over networks designed when text e-mail for researchers was the state of the art.

Participants at this week's O'Reilly peer-to-peer conference identified limitations in the Web address system as one of the key problems holding back the next steps in the Net's growth.

"The Net is broken, fundamentally broken at this point," said Ray Ozzie, chief executive of Groove Networks and the creator of Lotus Notes, citing barriers such as firewalls and scarce Net address availability.

"The Net doesn't work anymore like it was intended," Ozzie said during a panel discussion at the conference.

A few new infrastructure companies have achieved wildfire success, while others are still struggling for wide adoption. Akamai, which speeds Net downloads by storing Web page content inside networks physically close to computer users, has sparked a whole new category of infrastructure services, for example.

XDegrees itself is tying together several separate threads in the Net's development. Much of its technology is based on the peer-to-peer network models that Napster's success has brought to the fore.

The company's CEO refers to the myriad files, software programs and other bits of data that live on computers and networks, but which don't have Web addresses today, as the Net's "dark matter." The company's technology brings that matter to light, allowing access to it in much more efficient ways than today's daisy chains of Web servers, application servers and databases, he contends.

One powerful sidelight of this new model for Web addresses is the potential to eliminate broken links or vanishing Web pages--a feature of the Net familiar to many frustrated surfers.

Under the XDegrees system, multiple copies of files, content pages or applications could be kept in different places in the network. Similar to what happens with Akamai's content-delivery service, trying to reach this file through a URL would bring the surfer to the closest, or fastest, version.

Moving this file or its copies around the network, or between machines, wouldn't create broken Web links. Instead of pointing to a specific, inflexible address on a specific server, the Web address would point more generally to the file itself, wherever it might be.

Instructions on how to find wandering data would be relayed to an XDegrees server analogous to the machines that now translate Web addresses like www.news.com into a series of numbers computers understand.

"Right now it's like if every time we moved houses, we had to change our names," Tanne said. "That may work for the FBI protection program, but not for the Internet. We want to make files visible, not invisible."

So who uses it, anyway?
The task facing Tanne and his company is to persuade people to use these ideas. Even the best Net technology will be useless unless plugged into a network, and it can't genuinely be tested except in the wilds of everyday Net use.

A few other companies have suggested other ideas for expanding the domain name system, such as substituting keywords along the lines of "dogs" or "Nike" for traditional Web addresses. Such systems have been slow to catch on, however.

Tanne hopes to sell his company's service to big companies offering software services over the Net. He's also targeting Internet service providers, which he says could save on bandwidth costs if parts of traffic-intensive offerings like a music download subscription service could be moved inside their networks for those ISP's customers.

To some extent, he'll be facing competition from other competing innovations. Akamai offers Net-speeding services, and similar companies such as Edgix have business models that already target ISPs. Most of the big software companies, notably Sun and Microsoft, are already well advanced along their plans for distributing their wares online using existing or internally built infrastructure tools.

Nevertheless, the idea is creating a buzz.

"There are a number of restricting ideas that the Net has grown up with, that to move forward will need to be blown away by something," Hurwitz Group's Quinn said. "This could be one of those things."