面向查询集覆盖的物化视图选择算法

Materialized View Selection Algorithm for Query Set Covering

为了解决大容量物理存储条件下数据仓库的物化视图选择问题,提出一种面向查询集覆盖的物化视图选择算法.首先给出了一些概念和定义,然后从视图集的多维数据格中抽取和裁剪出候选视图集,并定义视图物化的效益模型,最后在存储容量的限制下逐步淘汰收益最小的应答查询的冗余视图,得到覆盖所有查询的最优物化视图集.实验结果表明,该算法在较大物理存储条件下的物化视图选择效率优于以往算法,且能够消除物化视图在应答查询时存在的时延"抖动"现象,应答用户查询的平均时间也大为缩短.

In order to implement materialized view selection in large capacity physical storage conditions data warehouse, a materialized view selection algorithm for query set covering is proposed. Firstly some concepts and definitions is introduced, then the candidate view set is extracted and cut from data cube, the benefit model is defined, the minimum profits redundant view response query set is filtered one by one when view＇s capacity beyond storage limit at last, eventually the optimization materialized view set for query set covering is obtained. The experimental results show that the proposed algorithm can obtain better performance in large storage condition than the previous algorithm, eliminate the time delay ＂jitter＂ phenomenon when materialized view in response query set and the average response time delay is also greatly shortened.