Research On Incremental Materialized View Maintenance In Data Warehouse

Because of huge volumes of date in Data Warehouse, excuting complex queries always take too much time. The materialized view, being an effective method, can improve the query speed and response user's request in time by pre-compute the data. When the data source changed, the materialized view need maintenance in time to keep in according with the data source. Aggregating queries are the key type of queries to data warehouse and on-line analysis processing,and directly affect the performance of the whole system. It is effective to preserve the results of aggregating queries to materialized views for improving the performance of aggregating queries.This paper prenents the properties of aggregating functions which is the key factor to the materialized views maintenance, and designs different ways for these functions to maintenan incrementally or near incrementally. This is divided into three parts for the research on incremental materialized view maintenance, according to the property of aggregating functions. The first part of research is on incremental maintenance for distributive function and algebraic function. Based on our maintenance model, we can incrementally compute many comlex and valuable functions, or user-define functions. The second part is on maintenance for ranking function of holistic function。With the help of assistant space, the method mentioned in the paper can compute ranking function nearly incrementally. The third part is on maintenance for median function of holistic function. With the cost of space, the paper puts forward a way to improve the view maintenance ability, just using the assistant measure set and moving window technology.