P2P plans could create speedier Net

Taking a page from peer-to-peer technologies popularized by Napster, these companies are looking at the ordinary desktop computer as a stand-in to perform some of the same duties now managed by massive banks of Web servers.

Despite peer-to-peer's proven success in Napster with content distribution, however, analysts are wary of security and reliability issues of what seems a fragile, unpredictable Net.

"Clearly there's value in being able to cut costs," said Joel Yaffe, an analyst with the Giga Information Group, a market research firm. But desktop computers are "so unpredictable...It would put content publishers in a precarious position vis-a-vis their infrastructure."

The system could work like this: A person wants to download a song. System software checks to find the fastest online connection. If a PC in the same building has the song and is close, the software connects with that PC and downloads the song. But if no other PC is available or close, the software instead points to a server out on the public Internet.

Start-up Kontiki is backed by many of the same people who originally created and ran Internet browser software company Netscape Communications, including Netscape founder Marc Andreessen. Red Swoosh draws on the talents of the core group of programmers who created what was once one of the most successful file-swapping services on the Net.

Both declined to publicly comment or offer many details, as they are officially in "stealth" mode for a few more weeks.

The ideas behind the start-ups' strategy have evolved as threads of peer-to-peer technology and content delivery research have gradually merged.

Akamai Technologies, as well as other content distribution companies, have built their businesses on pushing Net content--whether it be regular Web pages, individual images, or streaming media--as physically close to as many people as possible. With servers scattered within many ISP networks, Web surfers can reach content faster--easing pressure on network bottlenecks that have traditionally slowed Web connections to a crawl.

At the same time, peer-to-peer software developers have been looking for better ways to allow PCs to talk to each other while keeping network traffic to a minimum.

One good example is Napster's work with universities. Many colleges banned the use of the service because it used up so much of the school's network. With Napster's help, some universities figured out ways to route songs so that students were instead downloading songs from people at the same university--keeping the traffic within the same network.

Not without problems
The ideas put forth by Kontiki and Red Swoosh are not without controversy, and have certainly drawn considerable skepticism in financial and established networking circles.

Bob Gilles, CEO of QuMatrix, which is applying a similar peer-to-peer distribution idea inside individual companies, says venture capitalists told him to forget about applying the idea on the Internet.

Forget "this Internet thing," he said potential funders told him as he described his original idea. "Why would everyone allow anyone to get on their hard drive?"

With funding from a consortium of private Saudi Arabian investors, Gilles' company is now focusing solely on the corporate internal network market, with a first trial project scheduled at Citibank.

Executives at Akamai, one of the leading companies in the content delivery market, also are skeptical about bringing individual personal computers into what need to be industrial-strength networks.

"Security and reliability concerns mean we need to have control of the infrastructure for the bulk of our services," said Kieran Taylor, director of product marketing at Akamai. That would make adding ordinary computers "a bit of a challenge."

The companies aiming in this direction are not inexperienced. The idea has already been endorsed by RealNetworks, which is building a rudimentary version of the networking model into its MusicNet music subscription service. That will let people download songs from other subscribers if those subscribers' connections are faster than a central server.

Making it work
The start-ups too are mixing desktop computers with central servers. Each is building software that can identify where files are located and send requests for files to the cheapest, fastest place to find them, whether this is a nearby PC or a server inside the network.

Kontiki, which was known as Zodiac Networks until late last week, is funded by the Barksdale Group, Benchmark Capital and a handful of former Netscape veterans. Andreessen is involved as a technical adviser, and the company is led by Andreessen's Netscape colleague Mike Homer. The company declined official comment for this story.

However, sources close to that company say it is readying a public launch sometime in the next several weeks. It will provide software that can be bundled with a piece of content--a song or document, for example--that will then give individual computers the ability to send that content to others who are looking for it. Users would have to give their permission, sources said.

Red Swoosh, staffed by a group of engineers who previously built one of the most high-profile peer-to-peer systems, is pursuing much the same business model. With a product that sources say will be released in a public "beta" test version by late next month, they're hard on Kontiki's heels.

"Now content companies will be able to deliver 10 times more content to users at the same cost," said one source close to the company. "You could let a million-dollar company do the same things only a billion-dollar company could do before."

Executives at Red Swoosh also declined to comment on their plans.

The trick for these companies--or their customers--will be in drawing enough people into the networks to make the ideas worthwhile. As any veteran of file-swapping knows, peer-to-peer services are best when many people are on the network, so there are many places to find a given file.

Analysts say this could be a difficult proposition.

"Somebody's paying for this somewhere," Yaffe said. "You're just offloading the (bandwidth and storage) costs to the consumer."

But the start-ups believe that media companies can persuade their customers to offer up their computers--or TV set-top boxes, in the future--in return for lower subscription rates or access to content they wouldn't get otherwise. Inside businesses, a market QuMatrix and Kontiki each are approaching, the sell may be easier.

Ultimately the question about who participates will be an economic one, the companies hope.

"Content providers are going to give better pricing points for those customers that share," predicted a source close to Red Swoosh.