Without peers: How P2P will become mainstream

Headlines about Napster and Gnutella are seemingly everywhere and interest in peer-to-peer computing has become intense. But what does it mean beyond music swapping in IT departments around the world? Pia Heikkila takes a closer look...

Peer-to-peer (P2P) computing has been labelled the latest fad by some pundits and the press. Is that fair? The IT industry admittedly suffers from a constant hunger for the Next Big Thing, but real P2P enthusiasts claim recent Napster coverage has given the technology a bad name.

Napster and its counterparts get some of the blame for deterring any half-sensible IT director from utilising the technology but P2P computing is not all about swapping music across the internet.

P2P has its roots firmly in the early days of distributed computing. This model was established in 70s military and academic circles. At the same time, also gaining credibility was parallel computing, a way of using many computers to share CPU resources across a network to make connected machines function as one large supercomputer.

P2P applications can be divided into two basic categories. The first and probably the most famous one is based on file sharing, in which users access resources on other users' machines, as with Napster and Gnutella.

The other is distributed computing, where multiple computers are used to work on a single problem.

However, Robin Gear, analyst at Ovum, noted that P2P can be as simple as two or more servers talking to one another.

"Playing Doom on the internet is a good example of a P2P application," Gear said.
Explaining the technical necessities of P2P, he added: "The application must be designed to accommodate network connectivity patterns and temporary network addresses and all the peers in the network must be able to operate with a level of autonomy."

P2P based on a distributed computing model is basically about sharing a CPU's processing power across the network. It has attracted the attention of many organisations as large companies can have more than 10,000 PCs internally.

Dr Alex Wild, MD of Entropia, a company developing P2P tools for research purposes, said: "Many industries have CPU intensive problems including pharmaceutical, financial services, engineering, oil and gas. Companies there need supercomputers but one way around it is to utilise all the CPUs connected to the network," he said.

Another way companies can make the most of their PC armadas is to use machines in their offices in far away time zones when users are tucked up in bed.

One of the first internet-based distributed computing projects to utilise unused CPU power is called SETI@home. The project is based on a free screensaver, which users can download over the internet. Once installed, it runs a software program which analyses raw data collected by the Arecibo Observatory in Puerto Rico, sending it back to scientists once the analysis is completed.

The aim is to discover life on other planets and it claims to have nearly two million alien fans around the world. Many other similar projects have sprung up based on the SETI model, some urging users to donate their spare CPU power to cancer research and other good causes.

As lofty a goal as alien spotting may be, the P2P movement has yet to make a significant impact on the corporate IT world. Most IT managers are still coming to terms with other aspects of their networks. Adding another layer of complexity is enough to discourage most of them.

Richard Eckel, VP of marketing at P2P player Groove Networks, admitted companies are not consciously looking for peer-to-peer applications, but for the ability to utilise their networks better. "IT managers are seeking a type of technology which allows them to make the most of their IT investment," he said.

A survey published recently by Frost and Sullivan reveals Europe is far behind the US in the use of peer-to-peer. Robert McKellar, an analyst at Frost and Sullivan, said that Napster was a good way to educate users on how the technology works, but made selling the technology to IT managers difficult.

"Try to sell the image of decentralising information and it quickly sounds like [promoting] copyright violation. You just can't use the Napster image in a marketing message to IT managers," he said.

Sadly, the champions of P2P have to face the reality. As with any emerging technology, security is a high profile concern.

Ovum's Gear warned: "Because of lack of centralised resource, the files shared can circumvent corporate firewalls by establishing a direct connection between an external PC and an internal PC."

Vinny Gulletto, a P2P security specialist from Network Associates, said viruses are prone to the P2P environment: "It does not seem there is a lot of security built in to the available P2P tools, which means viruses can spread relatively easy."

Security concerns are not the only obstacle on the peer-to-peer path. Heavy file trading can also upset network traffic balance. Gulletto added that strict user guidelines are needed if brave IT managers are to take their first wobbly P2P steps. "If they are allowing users to have file sharing software they should have better security policies in place, as well as definitions of what is allowed to be visible on the network and what is not," he said.

Hype and negative publicity aside, the future of P2P still looks promising. It's simply because it can mean efficiencies that make very good business sense. Once the teething problems are overcome, the implementation of workaday P2P should become far easier than finding our extraterrestrial peers in the Milky Way.