Recent Replica Placement Algorithms in P2P Networks – A Review

Peer-to-Peer (P2P) systems provide a platform for large internet scale distributed applications with no dedicated infrastructure. In P2P systems, applications are developed by dividing them across individual systems termed peers which take the role of both client as well as server. Popularity of objects in such a system can change rapidly, which demands the need for a rapid and light weight content replication strategy which considers this dynamic popularity changes. While considering P2P in distributed file sharing applications, data availability has significant impact on the system's performance. In addition to optimized availability, the replica placement should guarantee reduced search and data access latency. It should be dynamically adaptable to instantaneous query arrival rate as well as dynamic membership of individual peers. Also, it should provide good performance with reduced number of control messages. Thus an intelligent placement of replicas considering various factors of the system always outperforms a random placement of replicas on random peers. This paper reviews and compares various recent replica placement algorithms for structured as well as unstructured P2P networks.