Data Placement Strategy For Data-intensive Applications In Cloud Storage System

Recently,with the coming of the "big data",as well as the continuous development of cloud storage and cloud computing,data-intensive applications have been received more and more attention.The huge amounts of data produced in the data intensive applications,has brought great challenges for the data placement in cloud storage system.On the cloud server,huge amounts of data access operations tend to be poor performance in the cloud storage system.Therefore,to design and optimize the data placement algorithm,so as to enhance the usage of cloud storage server and data access efficiency,is an important subject in the cloud storage system.In order to have greater distribution of the data placement in the cloud storage system,after a simple description of the data placement problem in the cloud storage system,which combined with the Lagrange relaxation method and related technology research both at home and abroad,method based on Lagrange relaxation decomposition was proposed.At the same time,as we all know,in order to improve the QoS of cloud users,replica is necessary.Users need data resource in the cloud storage system,so how to choose the best from multiple data copies to enhance the access speed is one of the key problems of data replica management.Then,in this article author research on data replica selection optimization problem,method of optimization strategy based on PSO was proposed.In this paper,the main work includes:(1)Elaborate the basic principle of Lagrangian relaxation algorithm in detail,As for data intensive application oriented data placement problem in cloud storage environment we established the data placement model.Under the precondition of three types of cost,which includes the server processing time,storage capacity and communication bandwidth,as well as integrated with Lagrangian relaxation decomposition principle,we put forward the data placement solution based on Lagrangian relaxation decomposition in the end.(2)In order to solve the data replica selection optimization problem in the layout management system,a copy of the replica selection of PSO based on PSO algorithm was proposed.Through the way of create particles on the access node,we reduced the time needed for replica selection.Using the grid simulator OptorSim for simulated experiment,we tend to have the simulation contrast test in terms of average job execution time,which compared the proposed Pso replica selection strategy with OptorSim own Simple copy of the choice of optimization strategy.All in all,this paper worked on the data placement and Replica selection optimization in cloud storage system for data intensive application,proposed the data layout algorithm based on Lagrangian relaxation decomposition model and replica selection optimization algorithm based on particle swarm optimization algorithm,the work in this paper has important reference value for further research.