OLAP cube partitioning based on association rules method

摘要

Partitioning is an optimization method used in Business intelligence (BI) systems to improve query and processing performances. That is why most of BI vendors integrate partitioning functionality in their solutions. However, they do not provide partitioning strategies which remain a serious challenge for BI administrators. Some works in the literature have proposed algorithms and strategies for Data warehouse partitioning. Nevertheless, most of them focused on the relational data warehouse partitioning and ignore the OLAP cubes although they are the first concerned by the user multidimensional queries. To deal with this, we propose in this paper a dynamic partitioning strategy for OLAP cubes based on the association rules algorithm. The first step in the proposal consists on analyzing the user queries for a specific period with a view to finding the frequent predicates itemsets. Afterwards, we use our proposed algorithm based on the association rules method to partition the data cube according to the frequent predicates itemsets obtained in the first step. Finally, we present a case study and experiences results to evaluate and validate our approach.