The Research And Implement Of Parallel Query Over Massive Data On Multi Database

With the rapid progress of computer networks, the total scale of inforamtion grows rapidly, more and more applications of the important bussiness start to use the Massive date-orenited Multi database. Those applications bring many problems to the traditional database technologies.On one hand, how to process and optimise the query on the Massive date-orenited Multi database is talked a lot in this field. There are many aggregation query on the Massive date-orenited Multi database, and it always be costy in the system performance. On the other hand, the faultcy coused by hardware is very high on such kind of system. the system always runs at 7×24 without stop, so it is really a big challege to improve the stability on the Massive date - orenited Multi database.This article based on backgroud of the Massive date - orenited Multi database, do some research on how to optimize the ggregation query and improve the system relaibility on the Massive date-orenited Multi database on the basis of the parael query technologies:1. in the Aggregation Query Optimization on Massive Data Based on Semantic Cache method, it accesses the Database to get the cache, and then refresh the cache information through the timing task. It will fall down the system's performance.2. To deal with the aggregation function, this article gives a query optimization way based on the model of hybrid DSMS, combining with the optimization method of semantic cache.This method gets the cache information by means of Query over an very small scale database, which obviously reduces the affects on the system's performance caused by the processes of getting the cache information through accessing the Database.3. the database system hot standby and mulit-computer cluseter technology are studied. A general strategy of database hot standby system proposed based on the thought of disk-mirror are designed and implement. This technology is ripe and cost less and the systems relaibility is greatly improve.4. all these research work is based on StarTPMonitor. On the base of our study, we develop system's function in paral query processing and aggregation query optimizaion and hot standby. The practical tests indicate that the function characteristics and the performance of this system is better than the target we desired.