Aggregate Query Rewriting in Multidimensional Databases

Abstract

An efficient query engine is certainly one of the most important components in data warehouses (also known as OLAP systems or multidimensional databases) and its efficiency is influenced by many other aspects, both logical (data model, policy of view materialization, etc.) and physical (multidimensional or relational storage, indexes, etc). As is evident, OLAP queries are often based on the usual metaphor of the data cube and the concepts of facts, measures and dimensions and, in contrast to conventional transactional environments, they require the classification and aggregation of enormous quantities of data. In spite of that, one of the fundamental requirements for these systems is the ability to perform multidimensional analyses in online response times. Since the evaluation from scratch of a typical OLAP aggregate query may require several hours of computation, this can only be achieved by pre-computing several queries, storing the answers permanently in the database and then reusing them in the query evaluation process. These pre-computed queries are commonly referred to as materialized views and the problem of evaluating a query by using (possibly only) these precomputed results is known as the problem of answering/rewriting queries using views. In this paper we briefly analyze the difference between query answering and query rewriting approach and why query rewriting is preferable in a data warehouse context. We also discuss the main techniques proposed in literature to rewrite aggregate multidimensional queries using materialized views.