Research On Configuring And Self Tuning Multiple Buffer Pools Of DBMS

Tuning the size of buffer area is the most important key in tuning configuration parameter of database management system. To raise the efficient use of buffer area, the whole buffer area is partitioned into a number of independent buffer pools. Database objects with different features are assigned to individual buffer pools. Also it can offer most appropriate configuration for a specific workload, which can reduce the time-consuming of i/os delays and improve throughput rating.After set up the model on configuring multiple buffer pools, the research on the multiple classify algorithm and greed size self-tuning algorithm has been done, and at last multiple buffer pools self-tuning system has been designed and carried out. Both algorithms are based on greed algorithm. During database running, this system can tune the buffer pools automatically, to stay the buffer pools in optimal solution and realize the self-tuning of database.Based on the model and algorithms, transaction analysis is conformed to the specific workload, to get the most effective parameter for self-tuning system. Single buffer pool self-tuning is presented in the multiple buffer pools self-tuning system firstly. Secondly we design the multiple classifying algorithms and present the traditional classifying algorithm. Finally we design the greed algorithms used in buffer pools size self-tuning and improve the algorithms efficiency. The system compares database performance with all kinds of self-tuning algorithms and configurations and displays all results of these situations. The result proves that the multiple buffer pools technique improves the database performance on the traditional buffer area, and so self-tuning system do.