An optimization problem in data cube system design

Abstract

In an OLAP system, we can use data cubes (precomputed multidimensional views of data) to support real-time queries. To reduce the maintenance cost, which is related to the number of cubes materiahzed, some cubes can be merged, but the resulting larger cubes will increase the response time of answering some queries. In order to satisfy the maintenance bound and response time bound given by the user, we may have to sacrifice some of the queries and not to take them into our consideration. The optimization problem in the data cube system design is to optimize an initial set of cubes such that the system can answer a maximum number of queries and satisfy the bounds. This is an NP-complete problem. Approximate algorithms Greedy Removing and 2-Greedy Merging are proposed. Experiments have been done on a census database and the results show that our approaich is both effective and efficient.