P2P is an acronym for Peer To Peer. Like VOIP, the term P2P describes the goal of a class of technologies, not any particular method, protocol, or individual technology. P2P has been the focus of much controversy since the first P2P network, Napster, was shut down on the grounds that it promoted copyright infringement. In general, Peer To Peer clients have since been very closely associated with copyright infringement, and because of this their use has been banned from many universities and ISP's. Of course, this also affects legal uses of the technology, such as the distribution of large files. As the only common, legal large files transferred via P2P are Linux distributions, companies interested in retarding Linux adaptation have generally shown favor in the repression of P2P usage.
P2P networks fall under one of two categories: structured and unstructured networks. Structured networks were the first type to develop, with the Napster model. In this model, one or more central servers store information about 'nodes', end users who both upload and download content to/from the network. This is similar to the more traditional FTP systems where one or more central servers are responsible for the distribution of content. However, unlike FTP, the central server in a structured P2P network is responsible only for the locating of content and the connection between provider and receiver. The actual transfer of data is handled by the nodes, thus greatly reducing the load on the central server. While structured Peer To Peer networks have excellent search capabilities, the use of central servers provides a point of compromise in the system. If the central servers are attacked or otherwise disabled, the entire network fails. This is how the RIAA shut down Napster: they forced the compromise of Napster's central server.
In comparison to structures networks, unstructured P2P networks are extremely resilient. The duty of the central servers is spread out amount the nodes. Thus, not only are the nodes transferring the actual files, they are also searching for files and establishing connections for unrelated nodes throughout the network. While in theory this may affect performance slightly (as the nodes take on extra responsibilities), in practice very few nodes are dedicated P2P machines. Thus, the added duties are only a very small fraction of the total work done by the machine, and is barely noticeable. With no single point of failure in the network, unstructured P2P networks are very difficult to neutralize through the use of legislation or even physical attack. However, there do remain attack vectors, such as file quality dilution and DDOS-style 'false traffic'. In a file quality attack on an unstructured P2P network, the attacker uploads many files with inaccurate or dangerous content, such as computer viruses. With no central governing body, there is no effective way to prevent the spread of these files, and the network becomes less attractive due to the low signal-to-noise ratio. In a DDOS-style attack on an unstructured P2P network, the attacker performs many high-load searches on the network, thereby disrupting the ability of regular users to search. Both these methods are currently in use by the RIAA to disrupt the use of currently active unstructured P2P networks such as Gnutella and Kazaa's FastTrack. Note that neither of these networks, nor any currently active P2P network, is a true unstructured network. They do share some properties of structured networks, such as central servers that help process searches, but are not vital to the operation of the network. Usually, these servers are physically located outside the legal jurisdictions of governments enforcing copyright protection.
Acronyms Software Terminology Questions