Index Technique In Datawarehouse

This thesis is the result of a study on Datawarehouse and Online Analytical Processing. To improve the capability of the system performance of OLAP, this thesis discusses multi-table join algorithm based on relations; The paper investigates popular index techniques and studys the bitmap index roundly. The contents of the thesis are briefed as follows:This paper studies the application of B-Tree index, bitmap index, index-organized table and B*-Tree cluster index in Datawarehouse and analyzes their pros and cons, while gives some advice on building and maintaining of several index methods of Datawarehouse. Bitmap index is an important Datawarehouse index technology, and executes the simplifying binary logic operation in order to optimize the query. The main features of bitmap index are high query efficiency, easy implement and good manipulation. But it is not adaptive to some environment, for example, it is not adaptive to range search which has more space price of index storing. This paper brings forward the bit-encoded index to realize range search. Verified by the experiment, the algorithm efficiently optimizes the procedure of query and achieves the expected goal.Multi-table join is one of the most time-consuming operations in the Online Analytical Processing based on relations. The traditional multi-table join processing method hasn't adapted to OLAP query at present. There are new multi-table join methods, such as multi-table join based on hash, multi-table joins based on sort which combine the star-structure feature. Comparing with the traditional multi-table join processing method, the performance has a remarkable superiority. This paper introduces a multi-table join algorithm based on bitmap index, which combines the bitmap and join methods. The algorithm can speed up OLAP query processing. Verified by the experiment, the algorithm efficiently achieves the expected goal. On the whole, the index technology of Datawarehouse is still a new topic, and every index has its merit, so we should think roundly to choose the index.