The Research Of Cloud Storage On Smart Grid Monitoring Data

In smart grid, by analyzing the grid condition data, we can real-timely monitor and predict power system status. Huge grid condition data has diversity of formats, not uniform, and some data require real-time processing. Faced with these problems, we need to use cloud storage technology to efficiently and quickly process and storage the mass of the grid monitoring data.Combined with MapReduce programming model of parallel data processing, BigTable,GFS and data storage technology, we propose a cloud storage prototype system of smart grid monitoring data,that describing in detail the overall design of cloud storage, cloud storage architecture, and also raise running processes and recovery strategy of cloud storage system, which builds a complete, efficient, reliable data storage and processing systems.This paper combines clustering algorithm and consistent hashing-aware data placement algorithm to distribute data. First, combine the factors of processor, memory, newwork speed and others to cluster the storage sets, and give priority to high-performance data server; Secondly, use the consistent hashing-aware data placement algorithm to distribute data to each server in the cluster.In order to meet the correlation between the data, convenience to access data, and to look for good network environment of efficient calculation migration, there is a need to re-optimize data distribution for the stored data. We make use of genetic algorithm to select the most reasonable method to optimize data distribution.Some experiments show that the proposed algorithm has a certain data distribution feasibility.Data queries has differrent efficiency because of different query sequences, and the efficiency among them has great distance. Moreover a special inquiry process of distributed data contributes to high query efficiency. Firstly, this paper compares the different query methods, showing the efficiency of the different query methods. Then, it optimize query commands by using algebraic optimization, improving query efficiency. Furthermore, it demonstrates the feasibility of a distributed data query. Finally, two multi-server collaborative inquiry procedures are given:iterative queries and recursive queries, and are compared.Scopes of this paper range from cloud storage prototype system of smart grid monitoring, to data balanced distribution and redistribution, and to the multiple servers distributed collaborative data query of distribution data. They are from the data storage to data query, that form a complete system.