Fragmentation Impact Analysis And Flash Memory Database Index Technology

Flash memory is a popular alternative for the design of storage systems because of its small size, shok-resistance, low-power consumption and non-volatile properties. Flash memory has been widely used in various embedded computing systems and portable devices such as MP3, memory cards, cellphone, PDA. The cost of flash memory is rapidly decreasing, while the storage capacity is rapidly increasing. This opens up the possibility to store bulk data on flash memeory rather than hard disk devices. The flash solid state disk / hybrid hard disk is popular used in personal computers and server systems. Since the application of flash memory, iphon, ipad and itouch become so popular. Flash memory management technology, especially for solving large-capacity data storage, flash-based database system, has become the focus of research.This paper describes the existing flash memory management technology, focusing on analysis of based on log-structured flash storage management. According to the different cost of all the merging operation types in the process of flash memory management, this paper analyze the impact of fragmentations on the cause of flash memory. On the basic of the reason analysis, padding the spacing is put forward, which changes the merging operation types. Through turning the fragmented writing into the sequential writing, this programe improve the storge performance of flash.This paper analyzes the main flash-based database system research directions, focusing on the indexing technique. The main content of this paper is how to efficiently implement the B+-Tree index structure on flash-based database system. Therefore, based on the physical characteristics of flash memory, combined with the existing flash memory management technology, this paper designs a modified B+-Tree index implementation layer, named MB+. This implementation layer, which is composed of Batch Process Buffer(BPB) and Leaf Node Header(LNH), reduces the number of write operations when the database updates, while improving the efficiency of flash-based database queries. Experiments and performance analysis show that the proposed program greatly improves the performance of the flash-based database.