Unbalanced replication by evaluating popularity over web scalable storage

Abstract

A peer-to-peer-based storage system has become a common way to alleviate network congestion and the limitation of storage space for providing file sharing, distributed computation, and instant messaging. Since the peer-to-peer technique has been used in file sharing, replications of a file is produced for a quick response and high availability. On the other hand, too much replications of a file may waste storage after the file become unpopular. The objective is to manage replications of file efficiently. In the proposed method, a popularity level is defined for evaluating the popularity of a file, used to determine the increasing or decreasing of the number of replications. The popularity level has three thresholds to determine replication of file. Thresholds are setting up by the 80–20 rule, and it changes dynamically overtime. We conclude that the popularity of file and its number of replication affect each other to use storage efficiently.