Research On Data Replication Management And Layout Optimization Strategy For Cloud Storage System

With the rapid development of cloud computer technology and the widespread application of the Internet,more and more user data are stored in the online cloud storage system.Data replica technology is an effective technical means to manage various data in the cloud storage system.It has the advantages of improving data access speed and data availability in the cloud storage system.Since replication management strategies has an important impact on cloud storage system performances,how to design replication management strategies is of great significance to achieve the best cloud storage management.In addition,as the scale of cloud storage systems continues to be expanded,energy efficiency problems of cloud storage systems are becoming more and more prominent.How to further optimize the layout of replicas to integrate resources and improve the effective utilization of energy in the cloud storage system is also of great practical significance for solving the energy efficiency problem of the cloud storage system.This thesis mainly conducts analysis and research on the data replication management and replica layout optimization strategy in the cloud storage system,focusing on the problems in the process of data replication management and replica layout optimization in the cloud storage system.The corresponding ideas and solutions are proposed from the aspects of theoretical modeling and algorithm optimization.The main work and innovations of this thesis are as follows:(1)Comprehensive consideration of multiple replication management problems in cloud storage systems including the cost of replication.These replication management problems are converted into a multi-objective optimization problem,and the performance indicators that need to be evaluated are mathematically modeled and defined as target space.(2)A replication management strategy based on the root growth algorithm(RMSRGA)for the cloud storage system is proposed.Firstly,RMSRGA simulates the self-similarity propagation model in the root growth algorithm.Secondly,it sets different growth strategies according to the area where different individuals are located in order to find the best individual in the target space as the most suitable replication management scheme.Finally,the experimental results show that RMSRGA can find a good balance among the number of copies,the position of the copy relationship and four performance indicators including mean failure probability,mean response time,the degree of load balancing and the cost of replication within a limited number of iterations.(3)The energy efficiency problem is transformed into a replica layout optimization problem.Three optimization goals are considered,including minimizing the amount of data nodes that meets the needs,maximizing the power of data migration savings and minimizing the difference between the CPU utilization and its optimal value of a data node.Besides,the optimizable objectives and constraints in the problem are mathematically modeled.(4)A replica layout optimization strategy based on greedy algorithm(RLOSGA)for the cloud storage system is proposed.In the process of optimizing the data layout,RLOSGA integrates resources to minimize the number of data nodes that meet the needs.In addition,the optimization of the location of each data is regarded as a sub-problem of the replica layout optimization problem.The greedy algorithm is used to get a more suitable replica layout within the conditions.The experimental results show that RLOSGA can more effectively improve the optimization goal and more success improve the effective utilization of energy in the cloud storage system in order to solve the energy efficiency problem in the cloud storage system.