Research On Hierarchical Storage Strategy Based On Multi-objective Optimization In Cloud Storage System

With the continuous development of network and information technology,the amount of data storage on the Internet has a great increase.With the single structure and low throughput,traditional storage system has been replaced by cloud storage,which solves the storage demand of huge data volume through clustered storage management technology,and greatly improves storage efficiency.However,simply being able to store data is not the only goal pursued by cloud storage technology today.In the actual cloud storage environment,data has different accessing situation at each stage of its information life cycle.Therefore,storing high-visibility data on higher-performance storage devices can improve cluster storage efficiency and improve users' experience.The tiered cloud storage technology can manage and make full use of the heterogeneous storage devices to improve the efficiency of cloud storage cluster,by providing suitable storage services for users with different data storage requirements.This paper deeply studies the existing hierarchical cloud storage technology,then proposes a multi-objective optimization hierarchical cloud storage scheme based on data pre-classification,which is used to realize hierarchical storage in cloud storage system.In order to organize the structure of the device and perceive the distribution of data heat,an applicability model based on device grading and data heat evaluation is proposed.This model organizes the cluster device through a three-tier hierarchical structure,calculates the device performance according to the device performance attribute,and establishes the device performance hierarchical structure of the system.In addition,the model evaluates data heat through data access,then the pre-classifies of data can be assorted by the stored value.Finally,the appropriate pre-selects storage area of each type of data can be determined according to the performance hierarchy of the device.In order to optimize the storage placement of data,this paper considers the efficiency of data placement by considering the key factors affecting the performance of cloud storage system.Taking the access time,access delay,migration cost and load change as the optimization goals,a hierarchical storage placement decision is established.The decision model is based on the improved multi-objective evolutionary algorithm,which weighs the pros and cons of these system performance impact factors,and makes the final storage placement decision for each data to obtain the optimal system performance.Then,in the cloud computing simulator CloudSim,this paper expands a series of functional structure related to hierarchical cloud storage,including the proposed hierarchical structure,data evaluation and storage pre-selection model,and the final storage placement decision scheme.Finally,after completing a series of simulation experiments,the hierarchical cloud storage scheme proposed in this paper can somhow reduce system consumption and improves data access rate.