ABSTRACT:
Ability to answer complex ad-hoc queries efficiently is a critical issue in data warehouse environment. The query execution time can be greatly reduced if the right index is built to facilitate fast access to data in the data warehouse. In this paper, we examine two most common indexes employ in data warehouse environment- B-tree index and Bitmap index. We evaluate the performance of the b-tree index against bitmap index on low and high cardinality columns using equality, range, aggregate and join queries. Our experiments results suggest that query processing time does not necessary depend on column cardinality, but depend largely on selectivity, that is the percentage of data retrieved by the select statement. In addition, we designed a concatenated bitmap index; our experiments show that concatenated bitmap index is more efficient in term of space, cost and particularly much faster than the corresponding composite b-tree for aggregate query processing in data warehouse environment.
Keywords:
Bitmap index, B-tree, Datawarehouse
