Traditional computer systems place resources on a centrally-managed server, which can then be accessed by client machines that connect to the central server (a "client-server" relationship). In contrast, Peer-to-Peer (P2P) technology allows client computers on a network to share their own resources (filestore, processing power, peripherals etc.) with other network-connected machines with little or no involvement of any central server. P2P software thus enables machines to act both as clients and servers. This powerful technology makes it possible for content to be distributed widely without the need for the central provision of large resources in terms of computing power, storage or in particular, network bandwidth. Unfortunately the technology is not without its problems, particularly for the network provider.
Some examples of current peer-to-peer resource-sharing software include: Skype, Spotify, Freenet, Gnutella, BitTorrent and Kazaa. Compiling and maintaining a complete list of such software would be virtually impossible.
Some definitions of "peer-to-peer" may include more traditional systems such as the Domain Name Service (DNS) or Usenet news (NNTP). For our purposes we restrict ourselves only to those systems where end-user systems are serving data direct to other such end-user systems, or relaying data via other end-user systems. Crucially the bulk of the communications on the system are not with one or more central servers, unlike such cases as an institutional NNTP server or the core global DNS infrastructure.
Up: Contents Previous: 1. Executive Summary Next: 3. Potential problems with peer-to-peer software
