Research Into Improved Method Of OLAP Execution Efficiency Based On Data Warehouse

Data warehouse be used to store large amount of historical data, OLAP applications involving large and complex queries of historical data in data warehouse, OLAP query optimization is the key to improve efficiency in the implementation. To set up Index is an important mean of data warehouse for query optimization. In the data warehouse system, the conventional index types are B tree index and bitmap index. B tree index is not suitable of multi-dimensional data storage,and for data warehouse applications are usually grouped with the polymerization conditions and the complexity of query, B tree index has not high efficiency. If the bitmap index for high-dimensional data base will take up much storage space,and update the bitmap index is also very high price. This article is mainly to improve the two indexs,present two new types of index structure which can reduce the query response time in order to improve efficiency in the implementation of OLAP.In this paper, by analyzing and researching, introduce the index structure of spatial databases into data warehouses, find out that QR tree index is a better data warehouse index structure than B-tree index. But QR tree also inherited defects of quadtree:When the depth is especially large,the speed will be slow, larger space will be occupied and regional sub-block will be not flexible.So its efficiency reduce with the increase of data. So based on the QR-tree Index,present an improved QR tree Index Structure-IQR tree index.IQR tree adds superstructure on the original structure of QR tree,which be used to store the data partition line rectangular. As a result,overlap scope of R-tree subtree will be greatly reduced, the efficiency of search will be significantly improved. Proved through experiments, IQR tree index is an ideal data warehouse index structure.For bitmap index disadvantage, this paper presents a new type of data warehouse index - Keyword Index. Through the partition of fact table, make the foreign keys of fact table to separated into separate tables. So that these foreign keys both as the data of fact table, but also as the index of fact table. This will not only save storage space, but also do not need to be updated on a separate index. An instance system of water supply company is designed and implemented with the keyword index, which proves the feasibility and validity of the Keyword Index introduced above.