Abstract: To make database systems always provide consistent high performance under various workload conditions, it is necessary to optimize database system settings. With the system becoming more complex and workloads becoming more fluctuating, it is very hard for DBA to quickly analyze performance data and optimize the system properly, and people resort to promising database system self-optimization techniques to solve the performance problem. A data mining based optimization scheme for lock table of database systems is presented. After training with performance data, a neural network becomes intelligent enough to predict system performance with newly provided configuration parameters. During system running, performance data are collected continuously for a rule engine, which choose the proper parameter of the lock table for adjusting, and the rule engine relies on the trained neural network to precisely provide the amount of adjustment. The selected parameter is adjusted according to the quantitative hints provided with the expectation that the system will perform better. The scheme is tested with TPC-C workload, the system’s throughput increases by about 16 percent.

Key words: database self-optimization, lock table, rule engine, neural network, predictor, data mining

摘要： 为了保证数据库系统在不同的负载情况下，始终提供强大的事务处理能力，必须对数据库系统进行性能优化。依赖于DBA，来分析性能数据，然后进行系统优化，在系统越来越复杂、负载持续波动的情况下是很困难的，数据库系统的自我优化，是很有前途的解决系统性能问题的技术。针对数据库锁表管理，使用基于轻量数据挖掘的优化方法，通过对性能数据的学习，建立一个能够根据锁表参数预测系统性能的神经网络预测器；在系统运行过程中，自我优化模块不断监控性能数据的变化，通过规则引擎选择需要优化的参数，利用预测器获得参数调整的幅度大小，完成参数设置，提高系统性能。实验证明，数据库系统性能获得近16%的提高。

关键词: 数据库自我优化, 锁表, 规则引擎, 神经网络, 预测器, 数据挖掘