一种查询反馈驱动的自适应统计信息管理方法

A method of query feedback-driven self-tuning statistics management

用一种查询反馈(query feedback,QF)驱动的方法对数据库管理系统的自适应统计信息管理进行了研究,设计了一个自适应统计信息管理系统的模型,开发了一个统计信息自动创建机制,研究了一种基于更新感知的统计信息更新方法,针对捕获的各种更新操作消息设计了有效的统计信息更新规则,并给出了关键算法。该方法利用执行器执行更新操作的反馈来更新统计信息,避免了对表数据的扫描,降低了计算量,提高了统计信息的更新效率和数据库系统的查询效率,同时又保证了统计信息的准确性,其有效性通过实验得到了证明。

The subject of self-tuning statistics management for database management system was studied using a query feedback （QF）-driven method. A self-tuning statistics management system model was designed, and a mechanism of automatic statistics creation was explored. A method of automatically updating statistics on top of perceiving data modification was presented, and the heuristic rules of statistics updating were designed. By using the feedback of executor to update statistics, this method avoids scanning data and reduces the cost of computing. Finally, the effectivity of this method was testified that it could improve the performance of updating statistics without losing the accuracy of statistics.