IBEM mentioned in article:
fortune.com
Sun Was Right: Why the Network Is What Matters
The real power of a computer is the combined power of all the other computers to which it is connected.
Napster is much more than a cool way for college students to steal music. The online music-trading service is the latest evidence of deep changes taking place in the design of the global computer networks on which the emerging digital economy is based.
In these networked days, the real power of a computer is less in the machine itself than in the combined power of all the other computers to which it is connected. Napster, for example, runs on commonplace desktop PCs, but because it links ten million such workaday machines to swap MP3 music files, it already, by some estimates, amounts to more computational force than all the servers at Yahoo. As the people at Sun Microsystems have been saying for years, the network is the computer. Its abilities depend not only on the total amount of intelligence but also on where in the network the intelligence lies and on how well it can share what it knows.
Distributed, interconnected digital intelligence is hardly new, of course, but it's becoming increasingly important. In its latest incarnation, it is threatening the recorded music industry and may soon threaten other forms of intellectual property as well. This phenomenon also promises to play a leading role in transforming the Internet into a mass medium and could one day alter the fundamental economics of the computer industry.
As a harbinger, consider the recent online broadcast by Victoria's Secret, the lingerie merchant known for its titillating catalogs. The Internet fashion show attracted more than a million hopeful viewers. Most saw whatever cleavage was on display, but for viewers in Houston and Chicago the show was a bust. They received only portions of the Webcast, and much of that in jerky images and halting sound. The reason is that millions of bits of data are needed to send high-quality video programs over the Internet, a process known as streaming, and general traffic congestion at the center of this global computer network almost always delays or loses some of those bits, resulting in poor sound or images for someone somewhere.
Faster transmission is one obvious solution, replacing the customary copper wires with conduits that can carry more information, such as the networks of through-the-air light beams planned by startups AirFiber of San Diego and TeraBeam Networks of Seattle, which could one day pump video at more than a billion bits a second.
More radical, and potentially more effective, approaches involve a redistribution of the intelligence that manages the streaming process, either increasing the computational power of the parts of the network that are closest to the viewers or bypassing the center of the Internet in favor of its edges. The latter approach has been used effectively by Akamai Technologies and Inktomi to speed the delivery of popular Web pages and banner advertising. Now it's being applied to streams of video data by Adero of Boston, iBeam of Sunnyvale, Calif., InfoLibria of Waltham, Mass., Microcast of New York, and a few other startups. The goal is to limit the number of so-called hops that streaming video must make from computer to computer through the Internet, thereby reducing delays and the chances of going astray.
By installing 2,500 servers around the country, Microcast, for example, believes it can deliver a million simultaneous video streams of 300 kilobits per second each. To avoid as much of the Internet as possible, and thereby its traffic delays, iBeam and Astrolink of Bethesda, Md., Cidera of Laurel, Md., and iSky of Englewood, Colo., are using satellites to transmit to outlying servers. Some hope eventually to install receiving dishes and video servers in office buildings, and perhaps even in residential ones. (I am an investor in iSky.)
Other approaches to networked computing are potentially even more far-reaching. In a way, they are similar to the research effort known as SETI, which has managed to persuade a few million volunteers to search for signs of life in outer space by using spare moments on their PCs to analyze radiotelescope data over the Net. For example, Napster, of San Mateo, Calif. (www.napster.com), connects an in principle unlimited number of personal computers, working together as peers. In contrast to customary Internet practice, any PC running Napster can act as a server to any other PC on the network. Freenet (freenet.sourceforge.net) and Gnutella (www.gnutella.wego.com) also use file-swapping software, which is being developed by ad hoc teams of volunteers to promote unrestricted and, their proponents hope, untraceable distribution of intellectual property of every sort, from spreadsheets to animation, music, and electronic books.
Following a similar design, a new company in Cambridge, Mass., Centrata (www.centrata.com), hopes to exploit the idle time of the estimated 350 million computers with access to the Internet. The company, which won a $10,000 runner-up prize in this year's new business competition at MIT, believes it can link unused machines over the Web, enabling owners to sell latent computational cycles and create what it calls "the world's largest hard drive and most powerful computer."
Now these are lofty claims, and obstacles are abundant. PCs, for example, are much more prone to crashing than the servers commonly used to provide information on the Internet, and typically aren't designed to operate all day, every day. In addition, because the traffic on Napster is virtually unmanaged, it has been slowing down Internet service on campuses across the nation, not speeding it up. In reality, there's little cooperation among the computers linked through Napster, which provides a central list of the MP3 recordings that have been stored on each PC in the network, but no indication of whether a specific recording can actually be downloaded from a specific machine when you happen to want it.
Still, paradoxically, the notion of networked communication will only grow in importance as broadband arrives. After all, broadband shares a key weakness with highways: Because wider roads inevitably attract more congestion, a driver in a hurry always needs to find better routes around the tie-ups. |
