Simulated Annealing for Materialized View Selection in Data Warehousing Environment

R. Derakhshan (Australia), F. Dehne (Canada), O. Korn, and B. Stantic (Australia)


Data warehousing, materialized view selection, query opti mization, simulated annealing


In order to facilitate query processing, the information con tained in data warehouses is typically stored as a set of materialized views. Deciding which views to materialize presents a considerable challenge. The task is to select from a very large search space a set of views that minimizes view maintenance and query processing costs. Heuristic methods have been employed to find near optimal solutions and recent genetic algorithms have significantly improved the quality of the obtained solutions. In this paper we intro duce a new approach for materialized view selection that is based on Simulated Annealing in conjunction with the use of a Multiple View Processing Plan (MVPP). Our experi ments show that our new method provides a further signifi cant improvement in the quality of the obtained set of mate rialized views, leading to a further significant improvement in query processing time and view maintenance costs for data warehousing systems.

Important Links:

Go Back