Self-tuning performance of database systems with neural network

Abstract

Performance self tuning in database systems is a challenge work since it is hard to identify tuning parameters and make a balance to choose proper configuration values for them. In this paper, we propose a neural network based algorithm for performance self-tuning. We first extract Automatic Workload Repository report automatically, and then identify key system performance parameters and performance indicators. We then use the collected data to construct a Neural Network model. Finally, we develop a self-tuning algorithm to tune these parameters. Experimental results for oracle database system in TPC-C workload environment show that the proposed method can dynamically improve the performance. © 2014 Springer International Publishing Switzerland.