Parallel multisource view maintenance
The VLDB Journal — The International Journal on Very Large Data Bases
| Hi-index | 0.00 |
A data warehouse is a global repository, which stores facts originating from multiple, possibly heterogeneous sources. The data warehouse stores these facts in a number of materialized views and keeping them up-to-date and consistent incrementally in response to the changes in the source data is referred to as Incremental View- Maintenance. In such a dynamic environment maintaining the consistency of warehouse is challenging because of the heterogeniety of data sources. This paper presents a scheme for maintaining data consistency at the warehouse. It does so by storing some additional information at the data sources as well as at the data warehouse. Our approach does require additional storage space, however, we argue that the effect of this extra storage requirement is negligible on the performance of the system. We achieve partial self maintainablity in the materialized views at the data warehouse by exploiting the already existing key dependencies in the views. We identify the problems in maintaining consistency and present an algorithm that minimizes blocking of sequential updates to be installed in a materialized view at the data warehouse and thus reduces the view-maintenance time.