Concurrent Warehouse Maintenance Without Compromising Session Consistency

To achieve acceptable query response times in data warehouse environments, the data have to be pre-aggregated according to the need of the applications and stored redundantly. These materialized views must be kept up to date, to assure their usability. In this paper we present an algorithm which improves the maintenance of materialized views in two ways. In the first place, the maintenance process is accelerated by supporting different maintenance strategies for individual views. Thus, views which do not need to be up to date can be maintained in a deferred or periodical way. In the second place, the algorithm provides session consistency, assuring that during an OLAP-session the underlying data do not change from the user's point of view. The algorithm facilitates high flexibility and performance of view maintenance without compromising session consistency.