Commentary: P2P needs compelling applications

The Internet application revolution has resulted in a migration of applications and data onto servers, dampening the growth of client-side performance requirements.

From Intel's perspective, peer-to-peer computing offers the potential of reigniting demand for more robust client systems--a trend that would clearly benefit the company. At the Intel Developer Forum on Wednesday, Intel Chief Technical Officer Pat Gelsinger commented hopefully that P2P could give a significant boost to the corporate PC market.

For individual consumers,

	See news story: Intel exec sees sales potential in P2P

P2P promises more control over their systems and a return to more "personal" computing. For businesses, however, P2P has yet to demonstrate a compelling value proposition.

P2P has at least three major drawbacks that will limit its adoption. First, it is costly. Second, it raises significant security and management issues. Third, and most important, it does not seem to have any compelling business applications.

What can go wrong
Many things can go wrong in a P2P scheme in a business setting. For instance, someone could turn his or her computer off in the middle of a process when others are attempting to access information. And as businesses move toward laptops, P2P systems with vital data may travel with their users and be disconnected from the office network, making the information unavailable to others. Server models eliminate these issues and the related costs to address them.

Although an ideal P2P model would store data on the client system of the consumers who "owns" the data, practical realities often require duplicates of the data to exist on other clients and servers. Moreover, the more heavily P2P is used, the more demand there will likely be for higher performance hardware and greater storage on client systems. As a result, the potential cost savings in reduced server spending is likely to be unrealized.

Finally, the increased energy costs--client systems would potentially need to be on 24 hours a day--would add to the total cost of a P2P solution. Even if the hardware in a P2P system is less expensive than a collection of servers, the real costs are in the design and administration of the system. These include not only the increased cost of IT resources to manage the environment, but the "soft" costs of end-user operations if individuals are forced to more actively manage their client systems.

Infrastructure developers carefully analyze the server resources and bandwidth needed by enterprise applications. P2P undercuts this careful design, generating significant uncertainty regarding bandwidth, server and distributed storage requirements. Vendors offering the first wave of P2P applications for business, such as Groove Networks, Ikimbo and Endeavors, have only started thinking about administrative controls, and it's going to be at least a year before they are likely to have credibility. In the meantime, these applications should be relegated to niches in workgroups and extranets.

Physical and logical security is also a major concern with P2P. Companies keep their servers locked up in the glass house and protect them with security software for a reason. Desktop computers in offices are much less secure physically, even with the front door to the office locked.

Linking all the desktop computers in an office through an internal P2P network also increases the exposure to hacker invasion tremendously--a hacker who can access one computer can quickly infect all the computers in the office through the P2P network. At least one worm has already been discovered on the Gnutella network. The most public and spectacular success with P2P may be attacks in which hackers take over other people's computers and have them all continually click on a specific Web site to try to overload and crash it.

What can go right
The biggest issue with P2P, however, is that few (if any) compelling enterprise applications are apparent. P2P is best used for repetitive, fairly simple tasks that can easily be divided among a large number of systems. SETI@home, in which scientists at the University of California enlisted thousands of home computer owners worldwide to help analyze huge amounts of radio telescope data from outer space for signs of intelligent life, is the best example--and by far the largest and longest-running P2P project to date.

SETI is an ideal distributed computing project because the data is easily parsed into chunks, scheduling is fairly easy since results do not have to be compiled into any specific order, and the system runs on a homogeneous computer population. P2P also has some potential as an adjunct to workgroup collaboration systems--for example, for highly distributed workgroups that are not tethered to a server. However, corporate financials, ERP (enterprise resource planning) and CRM (customer relationship management) applications would be nearly impossible to run over a distributed computing environment.

The problem is that computer tasks well suited to the capabilities and limitations of P2P are rare in the commercial world. The Gnutella search engine application, which searches the hard drives of all its subscribers for a specific file, may be useful to a workgroup, but few enterprises would be comfortable with the idea of search worms constantly going through their hard drives.

Consumers should track P2P developments for potential, targeted benefits in workgroup collaboration and simple, easily distributed applications. Some P2P tactical solutions may prove useful for ad hoc computing in the enterprise, mobile situations or some customer community situations. One feasible model, if executed correctly, would be a viral community that incorporates interactive marketing and community building. Also, there is a place for P2P in personal-area networks in which co-workers can connect PCs on the fly via wireless or infrared links to create workgroups in environments where there is no formal network. However, we do not see these as "killer" enterprise apps that will transform the corporate PC market.

If P2P is used, IT departments must closely examine the increased threat from hackers and proceed cautiously with applying P2P to even moderately complex enterprise applications. Any potential value from P2P must also be measured against the total cost, including upgrading client systems, managing the environment and providing adequate security.

Meta Group analysts Dale Kutnick, David Willis, David Cearley, Val Sribar and Mike Gotta contributed to this report.

Visit Metagroup.com for more analysis of key IT and e-business issues.

Entire contents, Copyright © 2001 Meta Group, Inc. All rights reserved.