Using Local Data to Maintain Views

Materialized views are used in a number of areas such as data warehousing, query optimization and mobile computing. A central issue in the application of materialized views is view maintenance, which is the issue of how to update a materialized view in response to changes in the source data. The standard techniques used by commercial database software for view maintenance are full view re-computation, and incremental view computation, whereby the view update is determined from the update and the source data. An alternative view maintenance technique is to update the view instance without accessing the source data, which is referred to as the self-maintainable (SM) technique. When it is applicable, the SM technique can provide efficiency advantages over alternative methods since it avoids having to compute potentially expensive joins operations in the case that the view contains joins. Also, it does not require the availability of all the source data, which can be an issue in distributed applications of materialized views. In this article, we develop a new technique for view maintenance. It is based on the SM approach but also allows access to the source relation being updated, but no other source data. We develop tests, to be executed at the update source as first-order queries, that are necessary and sufficient to determine if the view instance is SM in response to an update when the view is a conjunctive query. Also, if the views instance is SM in response to the update, we develop SQL procedures to update the view instances using only the view instance itself and possibly the updated source relation. We also report on the results of experiments on the efficiency of our SM technique, which show that it can be significantly faster than the incremental method. faster than the incremental method.