Analyzing and enhancing the resilience of structured peer-to-peer systems

In this paper, we propose an approach to analyze the resilience of structured peer-to-peer (P2P) systems under failures. The approach is Markov-chain based, and can be applied to systems with relatively stable size and uniform distribution of nodes. We apply our approach to several well-known structured P2P systems. We find that different system features (types of neighbors of nodes) in P2P systems have different impacts on their resilience against failures. Following this observation, we propose to add some extra neighbor(s) to CAN using small-world model principles to form a so-called CAN-SW system. We then apply the proposed approach to analyze its resilience. We find that the performance is improved significantly, particularly, in terms of the average path length. © 2004 Published by Elsevier Inc.