|
|
|
|
The peer-to-peer and Web services visions (as articulated in Microsoft's .NET and Sun's ONE strategies, for example) have much in common. In fact, "Web services" is a pragmatic subset of peer-to-peer computing. They envisage a decentralized overall architecture, with computers and applications given freedom to interact with each other with reduced centralized control. These decentralized architectures are gaining widespread recognition and will have a huge impact on the way information and technology is delivered and used. In anticipation, many of the early P2P infrastructure companies and standards bodies have been working to harmonize their products with Web services initiatives. Peer-to-peer in the enterprise The term "peer-to-peer" rocketed into the headlines in the second half of 2000. The application largely responsible for making it a household phrase was Napster, the infamous MP3 file-sharing application that has given the music industry so many problems. Other P2P applications that appeared at about the same time also helped to propel the term into the spotlight, such as SETI@home, Freenet, Gnutella and Jabber. What these P2P applications have in common is that they turn the user's computer from a passive consumer of Web information into an active resource within the Internet environment--making use of the often very powerful computers at the edge of the network, which usually lie idle or with large amounts of unused disk space. The success of these applications has spurred interest in the use of a P2P approach to business applications. Using the spare disk capacity and CPU cycles of an organisation's desktop PCs promises to bring greater flexibility and better return from a company's IT investments. In scientific and technical applications where substantia l processing power is needed and a "supercomputer" cannot be justified, a P2P approach--known as "grid computing"--can produce similar results (for certain problem types) as a supercomputer, at a fraction of the cost.All of this has been responsible for the creation of a host of start-ups seeking to build P2P applications in a variety of markets. It has also attracted the attention of some of the largest and most established of IT companies, including IBM, Intel, Sun and Microsoft. All of these companies are actively supporting P2P initiatives across a variety of areas. For example, Microsoft's commitment to peer-to-peer is shown by its recent $51 million investment in Groove Networks, a provider of P2P collaborative applications, a $1 million investment in the grid infrastructure provider Globus, and participation in the UK Government's GEODISE project. Sun has been working on Project JXTA and has acquired two specialist companies, Infrasearch and Gridware, to boost its efforts; IBM--another backer of Globus--plus HP and others have been publicly working on and promoting grid computing and other P2P efforts.A P2P application is one where a number of independent computers co-operate on a problem or provide a service, all sharing the load. Some observers believe that "true" P2P systems should not have a central co-ordinating server or management console, and that all work is done at the edge of the network. We would express this as: "any application or process that uses a distributed architecture and direct bi-directional communication between autonomous resources without central co-ordination and management." However, in practice, many other problems can be tackled if there is a central system tracking resource availability and parcelling out problems, like he most famous P2P applications, SETI@home and Napster. SETI@home uses a central system to divide up radio-telescope data for processing by Internet users' home PCs, when they are not being used for other purposes, and coordinates the results. Napster used a central directory to hold information on which MP3 music files were stored on its users' PCs--in effect, a yellow pages--which was made available for other users to download to their own machines. Therefore, a general definition is more useful for practical purposes: "any application or process that uses a distributed architecture and allows peers to provide and consume resources." What P2P applications can share We can deconstruct the definition by examining the types of things that P2P applications share. Peer-to-peer applications might share files, bandwidth, processing power, application components and/or raw data. These fall into three categories:
Centralized/decentralized applications Applications do not segregate neatly into categories of centralized and decentralized; instead most applications occupy the grey area in between. How centralized or decentralized an application is can be based on three criteria:
In addition, P2P applications will be easier to build, and more likely to make money, if software infrastructure companies can offer middleware to facilitate the creation and operation of some or all of these application types. This middleware should be based on agreed standards if it is to succeed. We therefore see an opportunity for P2P middleware companies. However, in general, revenue generation from such products will be indirect, not direct profit-generating line of business. The preceding article was excerpted from the report, "Peer-to-Peer Computing: Applications and Infrastructure," by Ovum, an analyst firm focused on converging technologies. Do you think P2P technology can deliver ROI on your enterprise's IT investments? TalkBack below, or e-mail us your thoughts. And, don't forget to vote in our P2P Quick Poll. |
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||
|