Data Warehouse Materialized View Consistency Maintenance

With the development of global economy, the application of distributed database becomes very important. More and more data is distributed in different areas naturally. For example, a company possibly has lots of filiales in whole globe, each of them maintaining its database. So data warehouse technology becomes more and more important. Usually, an enterprise information system is developed in a long period, consisting of a lot of isolated datadase, which run on different computer systems and DBMSs. That is to say, they are distributed, autonomous, and even heterogeneous. A data warehouse integrates data from distributed, autonomous, and even heterogeneous sources, provides a single uniform data store to support decision-making.Data warehouses contain large amounts of data. In fact, a data warehouse stores a lot of materialized views, which is available to user applications for querying and analysis. Since the data of a data warehouse is extracted from sources, they should be refreshed regularly (in each week or month) to make it consistent with the data of soueces. The maintence will be related to JOIN operation of several tables, if the data of a materialized view come from several data sources. Join operation is very important in the maintence of data warehouse. It directly affects the efficience of the maintence of data warehouse. Suppose the data of the view V come from table Rl, R2... Rn. If V is updated by rejoining these tables, it is costly because the result of rejoining includes V. Since we are only interested in the changed data, it is all right that we add the join result related to the changed data to V. This is called incremental join method.In this paper, we study incremental view maintence algorithms and propose a new distributed join method (incremental join method). Compared with the existing algorithms, our algorithm is easy to implement and has a good performance.The main works of this paper are:1 Study the main problems of data warehouse maintenance.2 Present an incremental join algorithm and prove its correctness.3 Comparing with the old algorithms, our method is easy to implement and has a goodperformance.