Optimizing a Data Warehouse using Evolutionary Computation

Abstract. A data warehouse stores huge amounts of data collected from multiple sources and enables users to query that data for analytical and reporting purposes. Data in a data warehouse can be represented as a multidimensional cube. Data warehouse queries tend to be very complex, thus their evaluation requires long hours. Precomputing a proper set of the queries (building subcubes) may significantly reduce the query execution time, though it requires additional storage space as well as maintenance time for updat-ing the subcubes. Creating suitable indexes on the subcubes may have additional impact on the query evaluation time. Proposed approach involves using evolutionary computation to select the set of subcubes and indexes that would minimize the query execution time, given a set of queries and available storage space limit.