Self-tuning Performance of Database Systems with Neural Network
Self-tuning Performance of Database Systems with Neural Network
复制标题
DOI：
10.1007/978-3-319-09333-8_1
复制
发表时间：
2014-08
期刊：
影响因子：
--
通讯作者：
Conghuan Zheng;Zuohua Ding;Jue-liang Hu
中科院分区：
文献类型：
--
作者：
Conghuan Zheng;Zuohua Ding;Jue-liang Hu
数据库系统的性能自调优是一项具有挑战性的工作，因为很难确定调优参数并为它们选择合适的配置值。在本文中，我们提出了一种基于神经网络的性能自调整算法。我们首先自动提取自动存储库报告，然后识别关键系统性能参数和性能指标。然后，我们使用收集的数据来构建神经网络模型。最后，我们开发了一个自调整算法来调整这些参数。在TPC-C负载环境下的Oracle数据库系统上的实验结果表明，该方法能够动态地提高性能。
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.