Solid-state disk page-level cache area management method

Abstract

The invention provides a solid-state disk page-level cache area management method. The method comprises the following steps of: dividing a solid-state disk page-level cache area into three parts: a hash index table cache area, a dirty page cache area and a clean page cache area, wherein the hash index table cache area is used for recording historical features of access of different data pages, thedirty page cache area is used for caching hot dirty pages, and the clean page cache area is used for caching hot clean pages; carrying out hot data recognition on a request data page by utilizing historical access feature information of a corresponding request on a hash table by adoption of a hot data recognition mechanism, and loading the recognized hot data page into a buffer area by combiningspatial local features of an access request; and finally, dynamically selecting proper data pages from a clean page cache queue and a dirty page cache queue to carry out replacement by synthesizing current reading / writing request access features and practical bottom reading / writing cost by adoption of a self-adaptive replacement mechanism when a data page replacement operation can be carried out in the buffer area. The method has favorable practicability and market prospect.

Description

technical field [0001] The invention belongs to the technical field of solid-state hard disk firmware optimization design, and in particular relates to a method for managing a page-level cache area of ​​a solid-state hard disk. Background technique [0002] The cache area is an indispensable part of the modern computer system and is widely used in storage systems, databases, network servers, file systems and operating systems. The traditional cache area management methods for mechanical hard disks are based on the assumption that the read and write operations of the underlying storage devices have the same delay. The number of write operations and the track seek time are the targets. However, the underlying storage medium of solid-state drives is flash memory, which is obviously different from traditional magnetic media. Therefore, traditional mechanical hard disk-oriented cache area management algorithms are difficult to apply in the design of solid-state drive cache areas...

AI Technical Summary

Problems solved by technology

However, both CFLRU and its latest improved algorithm ignore the dynamic read and write characteristics of the access load.

For example, unconditionally prioritizing the replacement of clean pages is not suitable for loads dominated by read characteristics

At the same time, neither CFLRU nor its improved algorithm makes full use of the local characteristics of data access, so it does not realize effective hot data page identification

Images