Abstract

A key element in all decision support systems is availability of sufficiently good and timely data to support the decision making process. Much research was, and is, devoted to data and information quality: attributes, assurance that quality data is used in the decision process, etc. In this paper we concentrate on a particular dimension of data availability and usage -the retrieval of data in a timely and decision enhancing manner. We propose to augment the decision support databases by an adaptive and efficient query cache. The cache contains snapshots of the decision support database, each being the answer to a recently invoked query. A snapshot can be reused by the originating user, or a different user, at a later time --provided the use of cached data leads to savings over the use of a new query, and these savings exceed the cost of using stale date. The proposed scheme is conceptually different from conventional data replication schemes. In data replication schemes the data items to be replicated and the protocols for concurrency control are defined at the system level. In our scheme the cache is populated dynamically and the snapshots it contains are refreshed only if the cost of using stale information is higher than cost of refreshing the snapshots. At the same time, users can still decide to refresh the stored snapshot, based on their own decision environment. Our scheme thus enhances the data retrievalprocess, while supporting a more efficient data retrieval at both the user level and the data warehouse level

Download

Share

COinS