Unstructured P2P models
-
Centralized P2P network: the network uses one server for indexing the files shared. Each time data is asked for, the server gives the user a list of the peers that have the asked data. User can then choose the peer he wants to download the data from. Server then facilitates the data transfer by removing itself from the process, so that these two peers now use a direct connection to each other, while still connecting to other peers via server. [illustration].
-
Hybrid P2P network: the network has a multi-central organization. For example, the network can be organized so that the peers with more bandwith get more tasks. Or, the network can have several nodes working as indexing servers, holding file name lists – again, a peer loads first the name file, and then will be able to connect directly to all the peers possessing this particular file.
-
Pure P2P network: the network consists only of peers, equal in strength. They have only one routing layer since there are no preferred nodes nor special infrastructure functions: each peer works as the others and so they're interchangeable. When connecting to the network, the peer connects to another peer, sending it a signal that it is alive. This other peer, in turn, gives the list of all the peers it is connected to, these peers then repeating the process, and so on. When ever data is desired, the user asks, not for a user who has the file, but for the file itself. Then each peer noticing, that they have this file, sends a signal back to inform that they have the wanted file, and the user searching for this file can start to download it from the peers that have it. This way of organizing has many benefits. The benefits to the data sharing may be the clearest: because the data is in many places at same time, the network is secure in terms of saving and providing data; since the peers are serving one another instead of all occupying one central server, the growth in the number of peers linked into the network doesn't slow it down, but makes it run faster, as there are more peers with wanted data. The same power of massive contribution can be used in production. [illustration]