Research On Data Consistency Maintenance In P2P Systems With Super Peers

With the P2P network development, more scalable,available and instantaneous to all kinds of resource is required in the P2P system. Replication is an effective way to improve system performance and data availability. With the P2P application transferring from read-only static files sharing to read-write dynamical files interacting,a efficient replications' consistency maintenance method is strongly required between frequently-updated files and their replicas in the P2P system. What's more,the method is required not only strong consistency but also fast propagation of update messages.After highly comprehend analysis of the way to create replicas and maintain the replicas consistency at present,this thesis presents a new method:Replications Index Tree based Consistency Maintenance in SuperPeers strategy (RITCM) to deal with the problems.In this method,according to the different resource,all SuperPeers' indexs are organized in many different complete binary trees which are used in consistency maintenance. With RITCM, this thesis makes further research on the stability and fault-tolerance.Finally,A experiment is made and the experimental results show that the Replications Index Tree based Consistency Maintenance in SuperPeers can largely reduce the number of the duplicated messages. Moreover,it can obviously improve the coverage.Meanwhile,comparing the NSRIT(No Sequence Replication Index Tree) strategy,the RITCM' propagation of update messages is faster.