Classification of Sparsity Patterns and Performance Evaluation in OLAP Systems (データベースシステム研究報告夏のデータベースワークショップ DBWS2002)

概要

論文の詳細を見る
OLAP performs the applications that interact with users and require fast response time over large amounts of data of data warehouse. Because these applications need a lot of multi-dimensional aggregation, it can take many hours or even days to run if executed directly on the raw data. Therefore, various kinds of precomputation techniques are proposed to handle this performance problem. However, sparse data of OLAP causes the data explosion problem in precomputation process and decreases the performance of OLAP In this paper, we discuss the sparsity of OLAP data and the performance bottleneck problem. As a fundamental research on sparsity handling problem, we define the sparsity patterns and develop an automated data generation program according to the sparsity patterns. We evaluate the performance of two OLAP products, MS SQL Server 2000 Analysis Services and Pilot DSS (Decision Support Suite) with the generated data. Focusing on the sparsity handling ability, evaluation results also are presented.
一般社団法人情報処理学会の論文
2002-07-17