The following text is copyright 2002 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.

The rest of peer-to-peer

By Scott Bradner

Peer-to_peer networking has developed a very bad reputation in the last year or so mostly because it is a term that has been applied to ad-hoc music distribution networks. This use of peer-to-peer technology has been attacked by the music industry as the reason for the recent drop in the sales of music CDs. I don’t suppose that continued high prices or a lack of music that people want to buy has anything to do with the drop. There are many other uses of the concept of peer-to-peer networking, which should not get overlooked in the fog created by the music industry's zeal to maintain outdated business models. One of these is GRID computing.

About a year ago I wrote a column about GRID computing (A blurry vision, Aug 13, 2001) in which I argued that they hype that painted the GRID as the "next phase of the Internet," as the New York Times put it, was over blown. I also said that the technology of distributed computing did have some significant uses, even if I did not think that it would be common that people will be sharing their local computing resources with people they do not know. I still hold the same view but a recent visit to the iGrid 2002 conference (www.igrid2002.org) showed me how far this type of peer-to-peer computing has come.

The conference was held in an extraordinarily well-connected science and technology center in Amsterdam. Connections included two 10 Gbps links to the US (one to New York and the other to Chicago) and many 2.5 Gbps connections to parts of The Netherlands and the rest of Europe. Over 400 attendees from 20 countries got to see more than two dozen live demonstrations and a full program of technology sessions. Most of the demonstrations were focused on the effective use of high speed networks and distributed high-performance computing with most of the rest focusing on the technology glue such as a security infrastructure needed to make this type of peer-to-peer computing workable.

To me the most emblematic demonstration was a real-time, interactive, 3-D work of art. "Art Flying In & Out of Space" by Jackie Matisee presents a collection of multi-colored long-tailed Japanese-style kites swaying in the wind. The work is presented in the CAVE, a walk-in virtual-reality environment in which images are projected on the walls and floor. The viewer wears special glasses to provide a 3-D experience. What makes this work of art a GRID demonstration is that the movement of each of the kites is calculated by a different computer and the computers are spread all over the network. It is a beautiful personification of distributed computing.

The demonstrations showed that the GRID technology is getting more mature, and with companies like the two-dozen sponsors of iGrid, including IBM and HP, pushing it, the technology has a bright future. But it still will complement, not replace, the Internet as we know it.

disclaimer: Historically Harvard's schools provided a good example of loosely-coupled distributed computing. Things seem to be changing, but whether it will replace Harvard as we know it will not be known for a while, meanwhile I express my own opinions