Self-tuning Performance Of Database Management System Based On Fuzzy Rules

Performance tuning of Database Management System (DBMS) is both complex andchallenging as it need to analyze the performance bottleneck of database management system,identify and set several key parameters from hundreds of parameters in the database managementsystem. In the traditional operation, the performance tuning of database management systemdepends on the Database Administrator’s (DBA) manual implement, its quality and the extent ofperformance enhancements completely relies on the professional skills and experience of thedatabase administrator. Thus lead to the performance tuning of database management system notonly has a lot of chance, and requires a lot of manpower and material resources to achieve them.Therefore, self-tuning performance of the database management system is a new technology thatis required in today’s high-tech society.In this paper, we present a method for self-tuning Performance of Database ManagementSystems based on fuzzy rules, i.e. self-tuning performance of database management systemsbased on Fuzzy rules. We first extracted the memory parameters values from the databasemanagement system memory parameter values, and extracted performance indicators values bygenerating AWR report, and then construct fuzzy rules library by analyzing the relationshipamong the memory parameter and performance indicator. Secondly, we present a newself-optimizing algorithm which analyzed the memory parameters values and performanceindicators values and compared with the related fuzzy rule match to improve the performance ofthe database management system by reconfiguration the memory parameters. In this paper, theexperiments on TPC-C workload and mainly divides into three parts: single parameter tuning,multi parameter tuning and two layers fuzzy rules tuning. The experimental results show that theproposed method is feasible and effective.The contributions of this paper are listed below.(1) This paper proposes a new method for the performance tuning of database managementsystem by reconfiguration memory parameters.(2) We first construct a fuzzy rule base based on the historical data, and then the performanceof the database management system can be self-adjusted through matching the fuzzy rules according to the run-time data.(3) This paper presents a new method of two layers of fuzzy rules, full consideration to theexternal environment affect the performance of the database management system, so as to makereasonable, effective and rapid adjustment.