Extended derivation cube based view materialization selection in distributed data warehouse

Abstract

View materialization is considered to be one of the most efficient ways to speed up decision support process and OLAP queries in data warehouse architecture. There are great varieties of research topics concerning view materialization, such as user query rewrite to transparently direct user query from base table to materialized views, or materialized views update as soon as base table changes, etc. Among most of these topics, a proper selection of views to be materialized is fundamental. While much research work has been done on view materialization selection in the central case, there are still no appropriate solutions to the problem of view selection in distributed data warehouse architecture, which is just the focus of this paper. We model the views in distributed warehouse nodes with derivation cube which is a concept widely used in central data warehouse, and make extensions in order to adapt it to distributed cases. Then, we propose a greedy-based selection algorithm under a storage cost constraint to perform selection process. Finally, a detailed experimental comparison is made to demonstrate the advantage of our solution over simply applying the central methods repeatedly on each warehouse nodes. © Springer-Verlag Berlin Heidelberg 2005.