Research And Improvements Of Database Performance Based On SAAS Platform

With the fast development of internet technology, the globalization of the knowledge economy, and the increasing competition environment forces companies must make full and effective use of modern information technology to business information. The traditional mode of business has long implementation cycles,high costs,inefficient, especially poor on small and medium enterprises in which case, resulting in the concept of SaaS,software as a service. SaaS brings hope for these enterprises to develop economic.The system of HiService is based on SaaS architecture which means multi-tenant, configurable, and extensible of the static data management. HiService is composite of workflow engine to implement business process modeling、execution and monitoring of automation. With the development of system, tenants are also increasing throughput, increasing the amount of data, the response from the server becomes slower and slower which the clients cannot stand waiting. The system efficiency is very low. A web server and a database server cannot support the application of the system.This paper is on the background of HiService system. Research and design the optimization of database performance based on the SaaS platform and build the system environment that composite of web server、cache server and database server clusters. On the first step, analysis current distributed storage strategy, then design the storage strategy of persistence layer for HiService system. On the second step, create correct index to improve index performance of database. Learn about Sphinx full-text search engine, design the method to improve fuzzy query performance, optimize databases executions. Using bind variables, database paging and other ways to enhance the efficiency of the implementation of improved SQL, combined with Web caching to improve the system performance further. The paper proposed parallel query algorithm based on the storage strategy of database. The database utilization is paralleled through the algorithm. Design and implement the data access layer transparent to the application layer by encapsulating details of persistence storage. Finally, test the improved system and analysis the test result.Through the experimental data, the improved method has bean verified feasibility and effectiveness, this research is the basis for better services for the system.