An Efficient Processing of Range-MIN/MAX Queries over Data Cube

Abstract On-Line Analytical Processing (OLAP) has become a crucial element of decision support systems. Since historical, summarized and consolidated data is used in OLAP, the concept of data cube is often used to provide multidimensional views for such information. Among range-aggregates that are typical operations over the data cube, we in this paper focus on efficient processing of range-MAX and range-MIN operations, which are commonly used but are still expensive even with other existing methods. We propose the concept of a maximal cover, which is an effective representation of data distribution information with respect to range-MAX and range-MIN processing. We show that the maximum and the minimum value of a given query range can be effectively computed by an appropriate maximal cover. Thus, the problem of processing range-MIN/MAX is transformed into the problem of finding an appropriate maximal cover. To speed up the search process, we propose the maximal cover network that is a search structure based on the containment relation between two maximal covers. Being different from the hierarchical tree proposed earlier, the search process in our maximal cover network completes when the first matching node is found. This property mainly contributes to the outperformance of our maximal cover network where the number of accessed nodes is significantly reduced. We compare the performance of the proposed maximal cover network with that of the hierarchical tree and show that maximal cover network works better.