The Research Of Data Warehouse Performance Optimization Technology Based On The Coupling Index And Materialized View

Index and materialized view are the physical structures which are used indatabase warehouse to improve the speed of queries. Based on the characteristics ofmaterialized views and indexes, the thesis proposed a optimization method that iscoupling index and materialized view to realize the optimization of query speed andstorage space efficiently shared, and the choice of candidate indexes and materializedviews is a data mining process. Finally, this paper used the cost model to evaluate thecoupling method and showed that the coupling scheme can greatly improve theperformance of a single index or materialized view through experimental analysis.The main tasks are as follows:The thesis firstly introduced the concept and the research of performanceoptimization of the data warehouse, including the definition of a data warehouse,features, the influencing factors of data warehouse performance optimization andseveral optimization methods, such as indexes and materialized views.Then, it described the theories of index commonly used in data warehouse andmaterialized view, including the definition of the index, bitmap indexes, the theoriesof join indexes and B-tree index and their application optimization; the definition andselection algorithm of materialized view, such as genetic algorithms, greedy andannealing algorithms.Based on the above theory, the thesis discussed in detail the idea of a newmethod of data warehouse performance optimization, it is data mining-basedwarehouse performance optimization. The thesis also described the steps ofoptimization method and candidate selection algorithm, by using data-miningtechniques to solve the problem of choice of candidate indexes and materialized view,as well as the cost model to measure the performance of index and materialized view.Lastly, the paper discussed the coupling method of index and materialized view,including candidate object selection of coupling method, data structures, and how tocreate an index on materialized view, by using cost model to measure the benefits ofviews materialized and indexed. There also listed the experiment charts of a singleindex, a single materialized view and the coupling case to compare the results, itverified the conclusion which the coupling method can greatly improve theperformance of system. In this paper, the coupling method about optimization of data warehouseperformance has a certain theoretical significance and application value.