Locating Data in (Small-World?) Peer-to-Peer Scientific Collaborations
Locating files based on their names is an essential mechanism for large-scale data sharing collaborations. A Peer-To-Peer (P2P) approach is preferable in many cases due to its ability to operate robustly in dynamic environments. Existing P2P location mechanisms focus on specific data sharing environments and, therefore, on specific requirements: in Gnutella, the emphasis is on easy sharing and fast file retrieval, with no guarantees that files will always be located. In Freenet, the emphasis is on ensuring anonymity. In contrast, systems such as CAN and Chord guarantee that files are always located, while accepting increased overhead for file insertion and removal.