Research On Techniques Of Load Balancing In P2P File Storage System

With the development of Internet applications and information technology, the scale of the data is growing. Traditional storage systems have not been adapted to the new storage requirements. The feature that DHT-based P2P networks have decentralized resource use and the decentralized self-organizing structure makes the file storage system which bases on DHT network good scalability and dynamic. And it has been a research hotspot at home and abroad. P2P file storage system is more scalable and redundant than the traditional storage system inherently. But it also brought the load-balancing issues.In this paper, load balancing issues of P2P file storage system have been researched on the basis of the predecessors'works. The results of this paper are as follows:First of all, the ID region which each node is responsible for is different, and node capabilities vary widely in DHT. To remedy this issue, a method based on virtual nodes division is proposed. A node performance model is established in P2P storage system, and then a load balancing mechanism is proposed. The weak performance nodes nearly in the physics network are divided into a virtual node to balance the system load in the mechanism.Secondly, the frequency of file accesses is different. A hot forecast and processing technology is proposed based on single exponential smoothing model. Use past query data and single exponential smoothing model to predict the future hot spots. The processing method is based on the user query behavior. The end-to-end replication strategy and light-load neighbor replication strategy effectively prevent and eliminate load imbalance caused by the hot spots.Finally, load balancing techniques for P2P storage systems has been researched and concluded in this paper. And methods are proposed to balance the load of in data distribution and data access. The experiments verify that the proposed scheme can effectively balance the load on network storage systems.