PartJoin: An Efficient Storage and Query Execution for Data Warehouses

The performance of OLAP queries can be improved drastically if the warehouse data is properly selected and indexed. The problems of selecting and materializing views and indexing data have been studied extensively in the data warehousing environment. On the other hand, data partitioning can also greatly increase the performance of queries. Data partitioning has advantage over data selection and indexing since the former one does not require additional storage requirement. In this paper, we show that it is beneficial to integrate the data partitioning and indexing (join indexes) techniques for improving the performance of data warehousing queries. We present a data warehouse tuning strategy, called PartJoin, that decomposes the fact and dimension tables of a star schema and then selects join indexes. This solution takes advantage of these two techniques, i.e., data partitioning and indexing. Finally, we present the results of an experimental evaluation that demonstrates the effectiveness of our strategy in reducing the query processing cost and providing an economical utilisation of the storage space.