Perspective of space and time based replica population organizing strategy in unstructured peer-to-peer networks

Abstract

File replication and replica management are effective strategies to ensure file availability and improve the efficiency of user access in peer-to-peer (P2P) networks. In this paper, inspired by the replica population model in population ecology, we propose a replica population organizing strategy for unstructured P2P networks. We model the activities of a P2P node from a time-and-space perspective, and model a file׳s availability from the perspective of node accessibility. The strategy automatically adjusts the relationship between the replica population and the environment, as well as the relationship between the replica population and the demand of users. Replica reproduction is well controlled by the strategy to adapt to the dynamic nature of P2P networks. We employ a push–pull approach, which enables replicas to change adaptively, both in their quantity and their distribution. Experimental results show that our strategy is able to adapt to the dynamic P2P environments. It effectively improves file availability, boosts retrieval efficiency, and meanwhile reduces the cost in network bandwidth.