Research On Invulnerability Methods Of Data Storage Networks

How to store data in storage systems,safely and reliably,is one of important challenges in data storage network.The distributed storage system,which has high scalability and good computing,has attracted more and more attentions.According to the target and deployment environment,we divided structures of the distributed storage system into P2 P networks and data center networks.They both have advantages of running and parallel computing,as well as could solve large-scale application problems.In this paper,the structures of data storage system were introduced by dividing them into traditional and distributed storage systems.Considering good scalability and data invulnerability in the distributed storage system,we mainly focused on its two major realization types: P2 P systems and data center systems.Then,we proposed the classification standard and the system model of storage optimization strategies in distributed systems.Based on detailed analysis of optimization strategies about above two systems,we studied their application scenarios and concluded their comprehensive performances.Finally,feasible combination mechanisms were proposed based on potential relationships between different categories.Based on above analysis of existing references,our research on the invulnerability of the data storage system was achieved by two parts as follows:The reduction of data survivability caused by the dynamic nature of P2 P systems,is one of challenges of the storage strategy design.Choosing proper replica nodes is a comparatively good way to improve the system fault tolerance,especially under high churn.Differing from existing references of the storage strategy,which always focus on specific P2 P systems,we proposed a storage method based on the zones partition-BRBZs,which mainly consisted of the copy distribution mechanism,the query mechanism and the consistency maintenance mechanism.It could be applied in different DHT systems.Then,compared with other strategies,we evaluated performances of the BRBZs by applying it in DHT systems with different routing protocols.Simulation results declare that the BRBZs has not only better data availability,higher query efficiency and lower searching fail-rate,but good scalability to be applied in systems with different sizes.Different from BRBZs optimizing the distributing mechanism in storage strategy,RSbo FMC was proposed based on the fragment matrix and the cache,which divided the optimization of storage strategy into replication and replacement strategies.The distributing mechanism in BRBZs partially depended on property of searching table of routing protocol,meaning it only applied different systems whose nodes had searching table.While in RSbo FMC,it was related to the zone number and fragment number after optimizing the replication strategy of initial storage state.In other words,it could be applied in different storage systems.Considering the reconstruction effectiveness and storage cost,the data partitioning mechanism based on fragment matrix and the cache mechanism were designed to optimize the replication strategy.Considering the load balance,the distributing mechanism,based on zone partition and opposite sequences,was designed to optimize the placement strategy.Simulation results show RSbo FMC outperforms other strategies in terms of the data reliability and the load balance.In addition,it has good scalability.