| A central problem faced in distributed data storage and in peer to peer applications is the efficient location of the node where the requested data is present. Currently popular peer to peer protocols such as Chord and Gnutella require enormous overhead. In Chord protocol data needs to be moved back and forth in order to stabilize the network and in Gnutella the requests flood the network. The above problems call for an efficient and reliable addressing scheme. The protocol proposed in this thesis (CHORELLA) can accommodate a maximum of 220 nodes in the network. It takes just one hop to find any file in the network and in case of node failures, it may take a maximum of as many hops as the number of mirrors or the replication rate in the network. The files are evenly distributed among the nodes using SHA-1 hashing function. |
