Research Of P2P Load Balancing Mechanism Based On E Xponential Smoothing Model

Structured peer-to-peer network (P2P) is one of new technologies in recent years, in the development of the current network, P2P technology plays a very important role with its excellent features and an equal sharing of ideas on the Internet. It has changed the traditional C/S mode access, making the network resource sharing more convenient between users. However, in the structure of the network, resources are accessed by different degree on the node, and the query request is not uniform sent by node, which has lead some nodes visits will exceed the actual processing capacity of node, will cause performance degradation of node. For the issue of access to the requesting node does not provide response services, affecting the quality of the entire network. So we need a reasonable peer network load algorithm to solve this problem.In this paper, a peer network load balancing studies conducted were analyzed. We study deeply the causes of hot issues, describes three solutions to load imbalance technology caused by hot issues. Through the three kinds of technology research, select copy and improved technology used in this project as a load-balancing technology. We can see from results of the experiment, this algorithm in resource utilization and response time nodes have been significantly improved.After analyzing the causes of hot resources, we proposed the hot prediction method for resources. By analyzing the exponential smoothing model, select an exponential smoothing model as predictive models. In this model, we use the network file access history information to predict the resource exceeds a threshold value. And according to a copy replication algorithm for resource creation, select the appropriate copy of the storage node placement. Through the analysis of experimental results, the method can effectively predict the hot resources, reducing node overloading where the hot spots are visited excessive.Through research of the load balancing, we propose the hot forecast and use copies of the cache to solve load imbalance problem.In the formation of hot spots, prediction model selection, the creation of copy’s numbers and the selection of cache nodes, we have carried out depth research. The final results also demonstrate this strategy of the effectiveness and operability.