Physical Network Based Replica Consistency Maintenance In P2P Systems

Replica technology is an efficient way to improve P2P system's scalability, fault tolerance, availability, and reduce the query response time. Replica, as an important part of P2P technology, has get more and more attention by people. In P2P network, that is, special resource has been distributed in duplicate nodes, these nodes constitute a replica network of the resource. How to maintain and manage the node has been called the replica management, it specifically include replica of creation, orientation, selection and replica consistency maintenance etc. Replica management in P2P network is closely related to the performance of P2P systems. If the replica cannot be managed well, it won't act in proper role and even reduce P2P system performance.Previous P2P replica research focuses on the search strategy and creating strategy. This static replica technology applies only to providing static data sharing P2P networks, along with the development of P2P technology. Today, data updating operation is more and more frequent, it is urgent to find an effective consistency maintenance algorithm to ensure the consistency of the replica.In this thesis, we first introduces the relevant knowledge of the replica management, and then analyses the present research at home and abroad, introduces the common strategy based on flooding consistency maintenance algorithm, and its improved algorithm, as well as the algorithm based on the data structure of the chain,the tree and the ring such as UPTRec SCOPE, PATCOM, etc. With the analysis of the advantages and disadvantages of these algorithm, we put up the Bandwidth and Delay based replica Consistency Maintenance Algorithm in P2P Systems named BDCMA, which takes into account the latency between nodes, and their bandwidth, creating a Bandwidth and Delay based replica Message propagation Tree BDMT updating the message, so that it can spread rapidly updating message and achieve replica consistency purposes. Through analysis, we can learn that with the scale of the system, the reliability in BDCMA algorithm increases, the coverage is high, redundancy is less, consistency maintenance time is short, etc.