Method for improving performance of solid state disk

Abstract

The invention discloses a method for improving the performance of a solid state disk. The method is characterized by comprising the steps of grouping commands according to logic addresses of writing instructions of a host in the solid state disk; dividing the writing instructions with the continuous logic addresses in the same group; and writing data of the writing instructions in the same group into a NAND flash with continuous physical addresses. After use of a command stream for management, the data of the continuous logic addresses is concentrated in the continuous physical addresses; evenif a user performs deletion and overwriting operations on the data subsequently, the old data is intensively invalid, and the whole block level is invalid with high probability, so that a garbage collection program processes invalid data blocks of the whole block disorderly; and the data can be reused after being directly erased, so that the data migration amount of garbage collection is reducedand unnecessary data writing is reduced.

Description

technical field [0001] The invention relates to solid-state hard disk control technology, in particular to a method for improving the performance of the solid-state hard disk. Background technique [0002] When the solid-state disk completes the user data writing, it usually does not distinguish the data, and only performs the writing operation according to the order in which the user write commands are submitted. However, the front-end operating system may be performing write operations on multiple logical space segments at the same time, which causes the original logically sequential user data to be stored in scattered physical spaces. When subsequent users access this continuous logical address , becomes a non-sequential read operation on the SSD side, affecting read performance. At the same time, as the user rewrites and deletes data, a block with more valid data pages and invalid data pages is formed in the SSD, which increases the workload of garbage collection and re...

AI Technical Summary

Problems solved by technology

However, the front-end operating system may be performing write operations on multiple logical space segments at the same time, which causes the original logically sequential user data to be stored in scattered physical spaces. When subsequent users access this continuous logical address , becomes a non-sequential read operation on the SSD side, affecting read performance

At the same time, as the user rewrites and deletes data, more blocks with valid data pages and invalid data pages are formed in the SSD, which increases the workload of garbage collection and reduces the service life of the SSD.

[0003] On the client side, there may be multiple threads writing data to the SSD at the same time, and the data written by each thread may be continuous, but on the SSD side, it is necessary to respond to the data writing of multiple threads at the same time, and each The priority of the thread is the same, which will make the original logical address continuous data writing cannot be stored continuously in the SSD, and the data of multiple continuous data command streams will be mixed together

like figure 1 As shown in the internal execution diagram of multiple commands flowing down the traditional SSD, the three threads T0, T1, and T2 simultaneously write data blocks with continuous logical addresses into the SSD. The SSD needs to respond to the three threads at the same time, so that the command nodes cannot be continuous inside the SSD. It becomes messy, and it is also messy and disorderly when stored on the NAND flash

For the operation of such a large data block, it is often centralized writing, centralized reading, and centralized deletion on the client side. Therefore, the above-mentioned phenomenon will cause two effects: First, due to the messy storage of physical addresses, subsequent operations on this The sequential reading of some large blocks of data becomes discontinuous when implemented at the bottom layer, which will affect the data reading performance

Second, due to the high probability of centralized deletion by users, the operations that could invalidate the entire block of data become scattered in each physical block. After a long time, the effective data in the disk will be scattered, increasing the pressure of garbage collection, and then affecting the life of the SSD.

Images