A Load Balancing Method under Zipf-Like Requests Distribution in DHT-Based P2P Network Systems
Yuanli Gu, Ling Chen, Keming Tang · 2009
DHT-based peer-to-peer systems have been proposed in the past few years. Scalability, fault tolerance and searching speed of the DHT based P2P systems have been greatly improved in contrast to the unstructured P2P systems. These DHT-base systems have the problem of load imbalance. We present a load balancing method taking into account Zipf-like requests distribution and the impact of physical networks. We propose routing table updating and hot resources caching algorithms to balance the system load. We advance two algorithms in our method. The first one is to remove the some links pointing to heavy peers and connect these links to light peers, this will reduce the routing load of the heavy load. The second one is to make the light peer to cache the hot resources stored in a heavy peer and reduce the request load of the heavy peer. Experiments and performance study shows that our method is effective.