![]() ![]() ![]() |
![]() |
|
|
![]() ![]() ![]() ![]() ![]() |
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Return to Data Warehousing and Active Databases Selection of materialized views and maintaining them is an important problem studied in literature. In this paper, we develop the notion of Partially Materialized Partitioned Views (PMPV) in which a materialized view is first partitioned and only a part of the view is materialized. The partitioning is done based on the workload of the database: the read-only queries and the update queries. The PMPV approach offers many advantages. The notion of PMPV is independent of the view selection algorithm and the algorithm used for answering the queries using the materialized views. The space occupied by the PMPVs will be less than the normal materialized views and hence saving the storage space. Since the sizes of PMPVs will less than the normal materialized views, the cost of processing queries using PMPVs might be less than that of using the normal materialized views. The number of updates that need to be considered while maintaining PMPVs will be less than that of the normal materialized views. We discuss the algorithms for selecting PMPVs and maintaining them. The experimental show that the PMPV approach offers advantages in terms of saving the storage space and decrease in the query processing cost at the expense of increase in the maintenance cost of the PMPVs. ![]() ©2006 Association for Computing Machinery |