Implementation of B+Tree Index Structure on JFFS2

• Main Authors: Chu,Heng-Hsing, 朱恆興
• Other Authors: Joseph Arul
• Format: Others
• Language: en_US
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/02142418543725234801

碩士 === 輔仁大學 === 資訊工程學系 === 97 === With the rapid development of flash memory, it is necessary to provide high speed read/write with a fast access to the data having great capacity. The technology also upgrades throughout the usage of flash memory as its fundamental basis. Journaling file system plays an important role in flash memory which is a methodology/creation to record journal entries so as to solve traditional defects of files within associated systems. The concept of its designing is to track down alternations from the file systems and record them onto the journals. The purpose of such a program is to prevent any serious loss or damages which might be done to the file system. Journaling Flash File System version 2 (JFFS2) is a branch of journaling file system which is derived from the flash memory. The integration of Memory Technology Device (MTD) and flash memory provides an effective management while reading/writing/erasing operations within files, and also performs better upon wear-leveling and garbage-collection under Linux processing environment. Problems occur while applying to a large amount of data, since JFFS2 was designed for flash memory which has small amount of data. If B-tree structure is adapted as the managing mechanism upon retrieve/store mechanism on the back-end, it would be time-consuming and less efficient. B+tree is a key alignment to form an index structure. In B+tree, unlike B-tree, there are ‘index node’ and ‘data node’. Before storing information to the ‘index node’, the indicated files would be kept on the ‘data node’. Furthermore, that information is stored on the left-hand side within the ‘data node’. Such interaction creates an index association to increase the chances of searching any kind of materials. This thesis mainly discusses JFFS2-the organization of back-end administration and the B-tree structure. One can easily comprehend the discussion of how to replace the original B-tree fundamental structure by using B+tree’s structures and how B+tree works to enhance performance and improve the efficiency. To explore all the research intension stated above, we need to adopt Linux as a complementing platform, and also the kernel parts from the Linux to upgrade, proceeds beta testing, and finally manipulating the JFFS2’s source codes. The results of experiments reveal that overall JFFS2 performs better under B+tree structure rather than under B-tree structure.