Research On Data Cache Prefetching And Replacement In Mobile Communication Supporting Platform

The rapid development and popularization of the mobile communication technology bring profound changes to people's work and life. At the same time, higher demands are placed on the mobile communication network services: higher quality of network coverage and faster response of network problems. Telecommunications Operators usually monitor the quality of the mobile communication network, detect network problems in time and solve network problems by mobile communication supporting platform.Existing mobile communication supporting platform, whose disk database is mostly divided into sub-database, make data-reading and data-storage complicated and cost a lot of time. What's more, the database of existing supporting platform fails to be optimized specially according to the features of network. Because of such many applications in supporting platform, the stability of the database is low. We introduce the Data Cache Middleware into traditional mobile communication supporting platform, to improve telecommunications operators' management and operation ability of network. In this thesis, we will research the data cache middleware as follows:1) According to heavy request on mobile communication supporting platform in daytime, we improved the traditional session Markov prefetching model. We present a new Markov model based on day granularity, which won't produce new request to the server on busy hours. Because of the wireless network optimization work, data request has two features: many periodically requests and new SQL that will be always made. Since these characters, we proposed a Dynamic Time Markov prefetching model. This method can predict the new SQL that has not appeared in train dataset, which contribute to the higher accuracy of F2 and lower bandwidths wastage.2) The size of SQL result set in mobile communication supporting platform can range from KB to GB. The huge difference leads to the classic GDSF algorithm's hypoesthesia of big files. Considering it, we proposed a new SIZE calculation formula combined with Taylor Series, enhancing GDSF performance in mobile communication supporting platform. We call this method as Adaptive Greedy Dual Size Frequency Taylor Series,AGDSF-TS. In AGDSF-TS, we have weakened SIZE calculation formula and modified the cache value function in hybrid access mode. The modifications of AGDSF-TS can reduce the average waiting time.3) Based on the above research work, the Data Cache Middleware(DCM) system is introduced into the mobile communication support platform. DCM have the functions of cache prefetching and cache replacement. We design a new data structure to transform the query results in database to JSON stored in Redis. DCM storage the query results in memory database, which achieves the goals of rapid data reading and avoids duplicated counting. Prefetching module and Replacement module of data cache middleware in mobile communication support platform could futher increase the cache effectiveness in mobile communication support platform, to implement the target of R/W Splitting. DCM improves the reliability and efficiency of the up-level applications' data reading in mobile communication support platform.