Research Of P2P Load Balancing Mechanism Based On Chord Procotol

Structured P2P network is a distributed network. The imbalanced needs of the storageresources and computer processing power heterogeneity may results that the visits of hot filesin some nodes will exceed its actual processing capacity in the peer-to-peer network.Degradation of nodes’ performance, request delay and network congestion are often caused bythis phenomenon of system’s load imbalances. This will affect the service quality of network.Dealing with P2P network’s load imbalance issues effectively is positively significant topromote the network performance and guarantee service quality.The article analyzes hotspot problems caused by some recent load balancing studies ofP2P network, researches the causes and the solutions of hotspot problems. After analyzing thepros and cons of existing solutions, the article selects an improved copy technology as a loadbalancing algorithm in this project.The article proposes the hotspot prediction method based on resource access for theproblem of load imbalance. By using historical data to predict recent visits of the node, thearticle copies files in advance of the possible visit hotspots and then alleviates load imbalanceproblem of the network. By comparing ARMA, BP and single-exponent smoothness models,the paper uses every model as predictive model to predict the access conditions of hot files.Through the analysis of experimental results, the method can predict effectively the hotresources, and reduce the problem of node overloading caused by hot files.The article selects the appropriate node putting the resource copies to avoidconcentrated access in connection with recent hot files. The paper conducts a thoroughresearch on the formation of hot files, the selection of prediction model, the creation of copy’snumbers and the selection of cache nodes. The paper’s algorithm improves significantly atresource’s utilization and nodes’ response time from results of the experiment.