A CUBE SIZE ESTIMATION STRATEGY OF DATA WAREHOUSE BASED ON SAMPLING

Size estimation of multidimensional views of data warehouse is a fundamental step for data cube design, warehouse storage planning, and query optimization. In this paper, a sampling-based cube size estimation strategy of data warehouse based on statistical theories is proposed. It applies different sampling size enlargement strategies to different kinds of views to ensure a satisfactory accuracy. Furthermore, It applies some heuristic rules in reducing the cube views according to the result of maximum estimation. An idea of sampling view pre-materialization is also proposed, which may result in a pre-materialized view set for generating an optimal estimation sequence of cube views that sharply reduces the total estimation time.