Abstract: The Multi Dimensional Expression(MDX) statements are often used in the process of multi-dimensional analysis of massive data. How to optimize the statements in order to boost the query speed is one of key issues when constructing the OLAP systems. By the predefined materialized views, the query statements can be rewritten and executed against one single table instead of the join of one fact table and several dimensional tables. The paper presents the thought of how to select the candidate materialized views and also the experimental results, which show that the boosting effect is more encouraging while dealing with more data via more complicated MDX statements.

Key words: materialized view, Multi Dimensional Expression(MDX), performance optimization

摘要： 多维表达式(MDX)语句多用于实现对海量数据的多维分析，如何对MDX语句的执行过程进行优化，进而提高查询速度是构建在线分析处理系统的一个难点。该文采用预建实例化视图的方法，使得原先基于事实表和维表的多表连接查询在经过查询重写后，可直接利用单一的候选实例化视图完成，从而大大加快了MDX语句的执行速度。阐述了选取候选实例化视图的基本思路，并给出实验结果。实验证实，数据量越大、MDX语句越复杂，性能提升的效果越明显。

关键词: 实例化视图, 多维表达式, 性能优化

CLC Number: 

• TP311.5