Research On Query Optimization In Dameng OLAP

Currently,Storing the large amount of historical data in the data warehouse for decision analysis,and data warehouse must support a variety of OLAP queries.But,the rapid expansion in the number of information,and these OLAP queries usually involve vast amounts of data,but also those data for a large number of complex operations,this is a very time-consuming process.Especially Dameng OLAP system is a Rolap system based on relational database,there are many deficiencies in its performance in dealing with multidimensional data,such as to conduct a large number of table join and aggretion opertions,so it is a serious problem to improve the query speed of Dameng OLAP system.In order to improve the system's performance,data cache technique and forecase management technology are applied to the Dameng OLAP system.and a cold start process brings aggregations from the relational database to cache,subsequently reducing the response time.Firstly, the problems of the current Dameng OLAP system are analyed,and the framework of the Dameng OLAP system based on management forecast,a cold start process and data cache technique is given out,then every part of this framework is described in detail.Secondly,three key techniques in the framework is studied:prediction management technology based on multi-dimensional data cold start process and cache management technology based on the result set.The system can capture knowledge about the query behavior of Dameng OLAP users by predict management system and learn the pattern of behavior that OLAP user executes MDX query,with the pattern of behavior ,it can predict which query results set is likely to use next and thus keep the most useful of them in the cache.In the system initialization and restarts,a cold start process is applied to load data into the cache system so that Dameng OLAP users can get alaways rapid response.at the same time the system design level cache management mechanism to cache OLAP user's query result set,select a reasonable cache object according to characteristics of Dameng OLAP system ,then give out a cache data management algorithms,and discusses the problem of cache consistency.Experiment shows that data cache technique and forecase management technology which are applied to the Dameng OLAP system can effectively solve the problem of performance bottlenect as the rapid growth of data warehouse,and the introduction of a cold start process when the system initialization process improves overall performance of the Dameng OLAP system.