Research On Fault Tolerant Storage Technology Of Cloud Storage System Oriented To Smart Grid

With the rapid development of smart grid construction, there are more and more collection points in the system. The collection of a medium-sized area can reach 20,000 to 100,000 records, and a large-scale area is facing 500 thousand to one million records. The amount of dada of a region expands from the current level of GB to TB level, smart grid requires storage systems for these massive data to store reliability. A solution of Cloud storage system is proposed to store massive data for smart grid, so it is proposed cloud storage system oriented to smart grid, with the following aspects of the job done.Firstly, this paper analyzes the problems faced by the smart grid and studies cloud storage system architecture. It is proposed that the data gathered from measurement system through the CIM modeling unified storage into cloud storage system. CRS code has fault-tolerant capability, space utilization and scalability, etc. The storage system is improved space utilization and data reliability through applying the erasure codes.Secondly, a algorithm called HGA(Heuristic greedy algorithm) is proposed to optimal calculation path, reduce XOR operation and computational complexity for data recovery, which applies CRS(Cauchy Reed-Solomon) code to cloud storage system HDFS and turns multiply operation of CRS coding to binary matrix multiplication operation. Then it takes the advantage of cloud storage system MapReduce programming framework for encoding and decoding by CRS to the line of parallelization operation.Then, this paper presents data placement algorithm based on the node’s credibility, node storage distance and the relative of load. When storing data for Smart Grid, cloud storage system requires some credibility value of data nodes, and node storage distance affects storing and reading data, while load balancing is an important indicator to overall cluster performance. The paper researches and analyzes data placement algorithm from credibility of the node, node storage distance and the relative of load, so that the data is stored with high fault tolerance and less access latency, reducing the probability of the same data simultaneously target multiple data blocks failure.Finally, this paper implements cloud storage system what it applies fault tolerant storage technology for smart grid. Using the laboratory equipments, it is designed to test the reliability of data storage, the efficiency of CRS encoding and decoding and the performance of placement algorithm.