Peer-to-Peer

Peer-to- Peer ( P2P) Connection (from English peer, equals ', ' coequal ') and computer - to-computer connection are synonymous terms for communication among peers, here based on a computer network. In some contexts, it is also called cross-communication.

In a pure peer -to-peer network, all computers are equal and can both receive services, as well as provide. In modern P2P networks, the network participants are divided, however, often depends on their qualifications in different groups take on specific tasks. Core component of all modern peer-to -peer architectures, which are mostly already implemented as an overlay network on the Internet, therefore, is a second internal overlay network, which normally consists of the best computers on the network and the organization of the other computers, and the shall provide the search function.

With the lookup operation peers can identify those peers in the network that are for a specific object identifier (Object ID) responsible. In this case, the responsibility of each individual object is assigned to at least one peer fixed, so one speaks of structured overlays. By means of the search operation to find the peers in the network according to objects that meet certain criteria ( such as file names or buddy - match). In this case there are for the objects in the P2P system does not mapping structure, so it is called unstructured overlays.

Once the peers that hold the objects sought, were identified in the P2P system, the file is directly ( in file sharing networks ), that is, peer to peer transfer. There are various distribution strategies which parts of the file should be downloaded from which peer, such as BitTorrent.

The contrast to peer-to- peer model is the client - server model. In this, a server provides a service and a client uses this service. In peer-to -peer networks, these roles are repealed. Each participant is a peer, because he can use a service alike and offer themselves.

Characterization of peer-to- peer systems

Typical but non- essential characteristics of peer-to -peer systems are:

  • Peers have a high heterogeneity in terms of bandwidth, processing power, time spent online, ... on.
  • The availability / connection quality of the peers can not be assumed ( " churn ").
  • Peers offer services and resources and accept services from other peers in claim (client-server functionality).
  • Services and resources can be shared among all the participating peers.
  • Peers form an overlay network and thus provide additional search / lookup functions.
  • Peers have a significant autonomy (through the provision of resources ).
  • The P2P system is self-organizing.
  • All other systems remain constant intact and not scaled.

( by: Steinmetz, Wehrle 2006)

Types of peer-to- peer systems

P2P systems can be classified into unstructured and structured P2P systems.

Unstructured P2P systems are subdivided again according to the nature of their construction. We distinguish:

  • Centralized P2P systems (eg Napster ), which require a central server to manage to function
  • Pure P2P systems without a central entity ( examples: Gnutella 0.4, Freenet ) A special type of a pure decentralized network forms the friend -to -friend - or web -of -trust network in which entertain no connections to unknown IP addresses be, but only links to friends (trusted friends) are established. (Example: Retroshare, InterFace (chat) )
  • Hybrid and Hierarchical P2P systems, which dynamically more central servers ("super nodes") to determine management (examples: 0.6 Gnutella, Gnutella2 (G2 ), JXTA )

Centralized and pure P2P systems is called systems are referred to as second generation systems of the first generation while decentralized systems. Systems that pass files over non- direct links, are third generation systems. See also detail the concept of file sharing.

Structured P2P systems often use a distributed hash table ( DHT). In structured systems, therefore, searches can be answered from a distributed index out.

Standardization and future

The future of peer-to -peer technology is mainly depend on whether it is possible to define a default - a kind of platform technology that makes it possible to set up other applications.

JXTA is such a standard, which was strongly supported by Sun Microsystems and open source. Sun set at the time the largest and most stable reference implementation ago.

Gnutella is another open standard that has been tested extensively, but so far almost exclusively used for file distribution and remote searching files.

It is also conceivable that the network transmission capacity will increase similar to the computing power in the PCs, so there is a possibility that a peer can not know the " on next " peer and that the sight of a peer about databases and other peers can continue to grow.

Client (computing) Friend-to-friend IBM Advanced Peer-to-Peer Networking
628890
de