Research On Cloud Storage Strategy Based On Erasure Code

With the development and popularity of cloud computing,cloud storage has become the main way of data storage.The explosive growth of data has led to an increase in the size and heterogeneity of nodes in cloud storage systems.Therefore,the reliability and I/O performance of data in nodes is strongly challenged.In order to prevent data loss caused by node failure,the system uses a specific recovery mechanism to protect the data.In this paper,we make in-depth research on the heterogeneity of nodes in the storage cluster and the multi-node concurrent repair problem,and the main job we do can be seen in below.(1)For the problem of high data storage cost,low reliability and insufficient node load balancing ability caused by the heterogeneity of nodes in the cluster,based on the segment sorting algorithm(FSSA),we improve the data block deployment scheme and propose a segmentation cross-deployment scheme.In this scheme,the data block deployment problem is first analyzed and mathematically modeled.Then the nodes are segmented according to their load condition,and some of them will be initially selected as the data block deployment candidate nodes.Finally,appropriate nodes for data deployment will be obtained after doing the selection and alternation operation of the data block deployment candidate nodes in each segment according to the idea of heuristic algorithm.Through the comparative experiments,the results show that this scheme can reduce the data storage cost,enhance the system load balancing ability and ensure the reliability of data storage at the same time.(2)For the problem of large system repair delay and low availability when multiple nodes fail in cloud storage,based on the multi-node concurrent repair algorithm(MCRA),we improve the node repair scheme and design distributed cross-repair scheme.In this scheme,the system's repair operation is performed in the replacement node and all the replacement nodes cooperatively and crossly repair the data to ensure that the data blocks required for repair are only transmitted once within the system,which solves the bottleneck of system repair and the large internal network bandwidth in the traditional repair scheme.Meanwhile,the repair trigger mechanism is adopted to avoid the problem of repair failure caused by the addition of damaged nodes during system repair,and the reliability of the system is increased.The simulation results show that DCRS has obvious effects in reducing system repair delay and increasing system availability.