The Research On Loading And Swapping Strategies For Main Memory Database

With the development of the memory technology, increased attention has been paid to the technology of main memory database in recent years. Much research has been carried out on how to build a practical main memory database system with high performance. However, little research has been devoted to the loading and swapping strategies for main memory database applied to SNS (Social Network Software) and web systems.This thesis presents a main memory database loading strategy for SNS at first. Then two main memory database swapping strategies are proposed. One is based on the Markov process, the other is based on expected pages mining technology.Firstly, community structure property was introduced into loading strategy for main memory database in SNS. GN algorithm is adopted to detect community structure in the strategy. Different from previous strategies, the strategy aggregates the data in close relation in a higher abstraction level. This strategy can be applied to any other systems based on social networks and main memory databases.Secondly, a swapping strategy based on Markov process for main memory database is proposed. Some attributes accessed simultaneously in relations are processed as an entity. The accessing process to these entities by transactions is proved to be a Markov process in the thesis. So the entities which are predicted not to be accessed in a given time can be swapped out from physical memory.Finally, web mining technology is applied to the swapping strategy for main memory database. A swapping strategy based on web log mining is proposed for web systems. The purpose of the strategy is to enhance the user experience with swapping the corresponding data sets into physical memory with the expected pages mining algorithm.The results of the experiments show that these strategies can improve the performance of the main memory database to a certain extent in the corresponding applications.