Research Of Replica Management Mechanism For Integration Of Cloud-P2P Computing

With the development and popularization of the Internet and social networks, traditional cloud storage systems have become increasingly unable to meet the storage of big data. As a result, This thesis proposes an idea for effective integration of cloud server resources and edge node resources. On one hand, cloud servers pressure can be shared. On the other hand, it can make fully use of the edge node resources. Replica management mechanism in Cloud-P2 P storage system has important impact on the system’s reliability, balance level and the quality of experience.In this thesis, we proposed one cloud data storage model. Besides that, thesis proposes a detailed description of the data storage method in Cloud-P2 P architecture. Meanwhile,one replica deployment strategy based on data availability and replica accessed frequency and remain capacity of host node, replica selection strategy based on replica response time’s prediction are also been proposed. Finally, a Cloud-P2 P streaming media system has also been designed which has applied the replica deployment strategy and the replica selection strategy proposed above. The following five points is a brief summary of the contents in this thesis.(1) Cloud-P2 P data storage model. This thesis defines the division of task of the two types of nodes in the model.The roles of node in management model are designed as three types: managing cloud node(SN), ordinary cloud node(CN), peer node(PN).(2) Cloud-P2 P data storage model. Based on Cloud-P2 P data storage model, we further give data storage method, including increasing of replica, decreasing of replica, modification of replica, reading-only access and removing of data.(3) Cloud-P2 P data replica deployment mechanism. We consider the data availability,dynamic replica access frequency and design the initialization and adjustment of replica numbers. Ensure the data availability and control of the number of replica to make it as little as possible. At the same time, the thesis also propose the method to select the replica deployment locations, which are determined rely on host node load balancing and system load balancing.(4) Cloud-P2 P data replica selection strategy. We integrate the factors that may affect the final response time and make the several different factors together into the actual replica response time. Through the statistical of replicas history of the response time to predict the response time of the replica accessed in future. And then select replica which owns the shortest response time as the best replica.(5) Cloud-P2 P streaming media system. Based on the above replica deployment and selection strategy we propose streaming system. We make a comparison of the solutions of the popular streaming media streaming systems. Put forward system architecture and program of performance optimization, that are methods of streaming data storage and sharing. At the last, we design block diagram and class diagram of the system.