File access method

Abstract

The invention provides a file access method used for carrying out indexing in a Flash database. The method comprises the steps that when AVL tree indexing nodes are upgraded, cached logs are generated in a log cache; when the log cache is fully written, the cached logs are combined with the node log set as an unit and written into a log storage area; a node log lookup table is set up, and the mapping relation between the nodes and a Flash page where the logs are located is recorded; log integration is carried out on the nodes and written into a node logic area. According to the method, the number of times for writing NANA Flash is reduced in the database upgrading process, and the upgrading efficiency of the index is ensured.

Description

technical field [0001] The invention relates to a file access method, in particular to a database file indexing method for an embedded system. Background technique [0002] NAND Flash memory has been widely used in various embedded systems, and has the advantages of high read and write performance, non-volatility, low power consumption, high density and good shock resistance. In embedded devices used for online environmental analysis, NAND is generally used to store index information sensed by sensor networks. At the same time, some embedded systems such as real-time environmental monitoring systems need to sense and store a large amount of data. As an efficient data management tool, the use of databases in embedded systems will become a trend. [0003] At present, mature databases are all used in disk storage systems. Most databases use a tree index structure to reduce access to disks when updating and retrieving database files. However, NAND Flash reads and writes in unit...

AI Technical Summary

Problems solved by technology

[0003] At present, mature databases are all used in disk storage systems. Most databases use a tree index structure to reduce access to disks when updating and retrieving database files. However, NAND Flash reads and writes in units of pages. Before writing, you must to erase

If the tree index structure of the traditional disk is directly applied to NAND Flash, for a large number of fine-grained file updates of the database, the block where the index is located must be erased and then rewritten, resulting in low update efficiency of the database file

[0004] Therefore, for the above-mentioned problems existing in related technologies, effective solutions have not yet been proposed

Images