Efficient materialized view maintenance in data warehouses

Abstract

In the data warehouse environment, the concept of a materialized view is common and important for efficient support of OLAP query processing. Materialized views are defined over several source relations. These materialized views need to be updated when source relations change. Since the propagation of updates to the views may impose a significant overhead, it is essential to update the ware-house views efficiently. Though various view maintenance strategies have been discussed in the past, the incremental maintenance of some kinds of views has not been sufficiently investigated. In the first part of this dissertation, we propose an efficient incremental view maintenance method for the select-project-join(SPJ) views. The proposed method minimizes the total cost of accessing source relations. We define the delta evaluation expression and a delta evaluation tree which are core concepts of our method. Then, a dynamic programming algorithm that can find the optimal delta evaluation tree is proposed. We also present various experimental results that show the usefulness and efficiency of our proposed method. In the second part, we propose an efficient incremental maintenance method for the data cubes. The data cube is an aggregation operator that computes group-bys for all possible combination of dimension attributes. When the number of the dimension attributes is n, the data cube computes $2^n$ group-bys, each of which is called a cuboid. To maintain a data cube incrementally, previous methods compute a delta cube, which represents the change of the data cube. We call a cuboid in a delta cube a delta cuboid. For a data cube with $2^n$ cuboids, a delta cube consists of $2^n$ delta cuboids. Thus, as the number of dimension attributes increases, the cost of computing a delta cube increases significantly. In the second part of this dissertation, we propose an incremental cube maintenance method that can maintain a data cube using only $^nC_[n/2]$ delta cuboids. As a resu...