Efficient incremental maintenance of data cubes

• Authors: Ki Yong Lee;Myoung Ho Kim
• Venue: VLDB '06 Proceedings of the 32nd international conference on Very large data bases
• Year: 2006

The data cube provides users with aggregated results that are group-bys for all possible combinations of dimension attributes. When the number of dimension attributes is n, the data cube computes 2n group-bys, each of which is called a cuboid. A data cube is often precomputed and stored as materialized views in data warehouses. The data cube needs to be updated when source relations change. The incremental maintenance of a data cube is to compute and propagate only changes of source relations rather than recompute the entire data cube from the source relations.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 2n cuboids, a delta cube consists of 2n delta cuboids. Thus, as the number of dimension attributes increases, the cost of computing the delta cube increases significantly. In this paper, we propose an incremental maintenance method for data cubes that can maintain a data cube by using only (n ⌈n/2⌉) delta cuboids. As a result, the cost of computing delta cuboids is substantially reduced. Through various experiments, we show the performance advantages of our method over the previous methods.