Research Of Replica In Structured P2P

P2P is in a dynamic environment,the network topology changes constantly,makes the establishment and maintenance of a P2P network and searching network resources efficiently become exceptionally complex.At present,the P2P distributed model has two basic types:unstructured and structured topology.Most unstructured P2P network uses a flooding mechanism for searching.It has a fairly high stability,but scalability and query efficiency is relatively low.The spread of control information consumes a great deal of bandwidth and very quickly cause network congestion and even network instability.The structure P2P networks mainly adopt distributed hash table(DHT) technology to effectively reduce the number of control information.It increases P2P network scalability,but it has poor stability.At the same time,the topology based on DHT is usually much more complex than unstructured network in maintenance and algorithm of repair.In this paper, we study the placement technology of replica in structured p2p,the placement technology of replica is the most common technology in a distributed data management System.Through increasing the replica,can be very effective to improve the availability and reliability of the data.Avoid peer which is widely distributed loses data, because of network or machine failures and other unpredictable factors. However , when the data upgrades, it is easy to cause replica inconsistencies. Especially when the upgrade in progress, and Responsible peer unreachable . When the peer re-join the network , will cause replica inconsistency.So a query in structed p2p is difficult to return latest replica.In order to solve the problem of replica inconsistency ,we presents a NTP-based Version Generator Service.It can generate monotonically increasing version number. Using version number,we can get the latest data.Compared with the paper[1],Improve the reliability of system .In addition,we make a simulation and test,and analysis effection on response time,when the number of nodes and the failure rate change.