A Load-balancing Approach for DHT-Based P2P Networks

Load balancing is a critical issue of the efficient operation of peer-to-peer (P2P) networks. Numerous proposals exist for load balancing in P2P networks. We propose an adaptive load-balancing solution for structured P2P systems. The approach aims to balance the request and routing load of the peers under biased request workloads, We balance the routing load through item balancing and the request load by caching the most popular keys along the lookup paths. These approach significantly improve the distribution of the load, and provide consequently better scalability. Our experiments shows that false positives rate is reasonably low with most workloads.