Implementation And Research On Design Of Main Memory Database

The memory database came into being because of technology advances, mature objective application conditions and needs of new applications. Memory database technology is a new database technology, and its design is that all or part of the database will be read into memory when the system is running, and it requires that all data operations are done in memory. At this time, I/O operation of the disk is no longer the bottleneck of memory database, so the goal of memory database design is to increase the speed access of the data of processor and space utilization in memory, while the data of external storage disk are only as a data backup of memory database. In general, the access speed of memory database is 10 to 20 times than the traditional physical databaseThe original design goal of memory database is to solve critical application systems in the telecommunications, finance and other areas. With the development of telecommunication services, in order to rapidly response to service request, telecom-class business support system loaded the more and more data into memory from the physical disk. Memory database plays an important role especially in real-time billing account services, online billing of real-time applications and embedded systems areas in the telecom business support system.This paper combine application needs of telecommunications to design and implement memory database. The first chapter describes the differences between the memory database and disk database, and highlights the historical research and development of memory database. We discuss the key technology of the memory database in the second chapter. We apply a shared memory mechanism which creates tables, fields and record structure. Then according to the characteristics of the memory database, it adopt index structure of the Hash table and T- tree to achieve the the balance between retrieval speed and storage efficiency. The third chapter designs data organization, data storage, data manipulation and row-level locking for data memory database. In addition, chapter four implements the core module of memory data, which include the main service process, data synchronization processthe data backup process and support flexible synchronization of disk databases (ORACLE, Sybase, DB2, etc.). Because this core module support data communication access mode based on SOCKET, direct connection (direct access) and etc, in connection with the host using direct access mode, you can avoid the overhead of inter-process communication, which make the application the same as using its own memory access memory database eve can improve greatly access efficiency. Besides, it can support the two-way star replication for specified table of multiple MMDB through providing replication of separate disaster storage recovery. Chapter three and chapter four that complete the core design and implementation for memory database system play an important role. In summary of the fifth chapter of system test include a performance test and high availability test.In summary of the sixth chapter presented conclusion of paper.