A Research On The Selection And Maintenance Of Materialized Views Used In Data Warehouse

Materialized view is a pre-action for those commonly used operations in OLAP. According to the possible queries of the OLAP, data warehouse can pre-process the fact table by selecting, projecting, joining and so on, which is called Materialized View. Because of the performance optimization from materialized view, it has become a hot research filed in the current data warehouse research. However, with the good performance,additional storage space and high cost of view maintenance are also resulted. In this paper, we made a thorough study and large number of experiments on materialized view selection and maintenance, and achieved the following results:we firstly introduce the background, application and the development of the materialized views at home and abroad, then cite a number of algorithms which are used in selection and maintenance of materialized views, after all we analysis the advantages and disadvantages of each method which were analyzed.A new selection algorithm which is based on multi-factor view selection is proposed; with our algorithm we can overcome the shortcoming of the existing algorithms including YKL and IMDVSA. With our experiments, we show that our algorithm have advantages in the actual cost.An algorithm focused on maintenance of materialized view is proposed, which can solve the problem of maintenance of the structure of materialized view, thus effectively improving the efficiency of materialized view maintenance, finally ensure the effectiveness and timeliness of the data.