A Case for Content Distribution in Peer-to-Peer Networks

In large scale peer-to-peer networks, it is impossible to perform a query request by visiting all peers. There are some works that try to find the location of resources probabilistically (i.e. non-deterministically). They all have used inefficient protocols for finding the probable location of peers who manage the resources. This paper presents a more efficient protocol that is proximity-aware in the sense that it is able to cache and replicate the popular queries proportional to distance latency. The protocol dictates that the farther the resources are located from the origin of a query, the more should be the probability of their replication in the caches of intermediate peers. We have validated the proposed distributed caching scheme by running it on a simulated peer-to-peer network using the well-known Gnutella system parameters. The simulation results show that the proximity-aware distributed caching can improve the efficiency of peer-to-peer resource location services.