Study On Storage And Index Technology Of Main Memory Database

In recent years, the MMDB (Main Memory Database) becomes a hot area of research in the database at home and abroad. The Main Memory Database plays an important role in the area of real-time application. With the price of the memory becoming lower, the integration of chips has improved more and more. It is feasible to store larger database in main memory. This makes main memory database (MMDB) becoming more practical. Because of its rapid data accessing ability, the main memory database is much more suitable for rapid response area and high-throughput applications than the disk database (DRDB). For those applications which need to be accomplished in a strict time period and which need to support concurrency of large amount of high-performance transaction processing platform. The main memory database is an ideal choice. This paper deeply analyzes the difference between main memory database and disk database, highlighting the characteristics of the main memory database. From aspects of the data structure and algorithm, this paper focuses on study of the memory database storage technology and cache memory optimization technology.In this paper, the first part introduces the application background and research significance of the main memory database. The main memory database applies to the real-time application which needs to be achieved in a strict time period. Then introduces the concept of memory database, definitions, comparison with the traditional disk databases, the advantages of the memory database. After that, we focus on the difficulties of research the main memory database. in the end, this chapter introduces the research status of MMDB. It is the base of following chapters. For the storage of MMDB, We enumerate several storage models which are used in MMDB, and describe the disadvantage of those models when they utilize the processor cache to improve performance. Then, we propose a data organization model called "Hybrid Storage Model", which improves cache performance by grouping together all values of each attribute within each page. After that, we study the structure of index of MMDB, and enumerate several indices based on trees, which are used in traditional MMDB systems. We demonstrate that in main-memory databases, the number of processor cache misses has a critical impact on the performance of the system. Based on the study of those indices' cache performance, we proposed a cache sensitive index called "Hash & Tree Tree (HT tree)", which combines hash method with tree. It gives attention to the effects of cache misses and TLB misses, and can gain better performance than conventional cache conscious indices. The final part is a summary of this thesis and expectation of the MMDB system, the direction of future development.