Research On Performance Optimization Technology Of Large Data Cloud Analysis Service

With the development of science and technology, the explosion of scientific data bring the great pressure for the management and analysis of scientific data. How to effectively manage and analysis these data become a huge problem. On the one hand, the model of scientific data is array, and the RDBMS’s data model is table, so it can not natural support array model. It combines the RDBMS and analytical software to analyze scientific data. However, this method will lead to expensive cost. Now there is an urgent hope a system that can manage and analyze scientific data meanwhile reduce the cost. Array database and cloud computing technology give the opportunities for management and analysis of scientific data. Array database regard the array as the first-class and can natural support scientific data’s storage and analysis. Cloud computing techonology integrates all resources to provide resources service, so it can reduce the cost.On account of this, combining the array database and the cloud computing platform to provide high performance scientific data analysis and management has important application value. This paper comprehensive analyze current array databases and a variety of cloud platforms and study the performance optimization techonology of science data management and analysis system. The main research contents of this paper as follows:(1) built the cloud platform Proxmox VE to support the analysis of cloud services environment;(2) designed and implemented a FASTDB prototype system for scientific data’s analysis services;(3) give two ways to evaluate the FASTDB, it can provides foundation for FASTDB’s performance optimization.(4) optimize the storage block segmentation strategy for FASTDB and implement the Cost-based optimizer based on array statistics. These two optimization methods have improved the management and analysis of scientific data in FASTDB.