CN109521961B

CN109521961B - Method and system for improving read-write performance of solid state disk

Info

Publication number: CN109521961B
Application number: CN201811346327.9A
Authority: CN
Inventors: 华荣
Original assignee: Shenzhen Union Memory Information System Co Ltd
Current assignee: Shenzhen Union Memory Information System Co Ltd
Priority date: 2018-11-13
Filing date: 2018-11-13
Publication date: 2021-10-26
Anticipated expiration: 2038-11-13
Also published as: CN109521961A

Abstract

The invention relates to a method for improving the read-write performance of a solid state disk and a system thereof; the method for improving the read-write performance of the solid state disk comprises the following steps: s1, acquiring the data newly written by the user; s2, the SSD stores the new data to the first space to be managed as hot data; s3, the SSD internally and periodically arranges the hot data of part of the first space to change the hot data into cold data and trigger data transfer; s4, the cold data is moved from the first space to the second space and stored. According to the invention, different data in the solid state disk are separately managed, the first space and the second space use different RAID modes for data protection, the RAID3 which does not influence the write performance is used for the first space for storing hot data, and the RAID5 which does not influence the read performance is used for the second space for storing cold data, so that the maximization of the whole read-write performance is obtained, and the requirement can be better met.

Description

Method and system for improving read-write performance of solid state disk

Technical Field

The invention relates to the technical field of solid state disk read-write performance, in particular to a method and a system for improving the read-write performance of a solid state disk.

Background

With the development of the solid state disk technology, an in-disk RAID protection function is introduced into most of firmware of the solid state disk, and when data loss occurs in a NAND Flash memory (NAND Flash), the RAID protection in the hard disk can recover the lost data through RAID error correction at a high probability, so that better in-disk data security is obtained.

However, in any current RAID protection principle, check data needs to be additionally generated, the check data (parity data) needs to be written into the NAND flash memory along with user valid data, which has a certain influence on write performance, and meanwhile, due to the existence of the check data in the hard disk, the subsequent data read operation may be influenced, which causes a decrease in read performance, so that the current solid state hard disk with the RAID function in the tape reel cannot meet the demand.

Disclosure of Invention

The invention aims to overcome the defects of the prior art and provides a method and a system for improving the read-write performance of a solid state disk.

In order to achieve the purpose, the invention adopts the following technical scheme:

a method for improving the read-write performance of a solid state disk comprises the following steps:

s1, acquiring the data newly written by the user;

s2, the SSD stores the new data to the first space to be managed as hot data;

s3, the SSD internally and periodically arranges the hot data of part of the first space to change the hot data into cold data and trigger data transfer;

s4, the cold data is moved from the first space to the second space and stored.

The further technical scheme is as follows: the hot data is data newly written by a user.

The further technical scheme is as follows: the cold data is effective data moved when the data in the SSD is stored and sorted.

The further technical scheme is as follows: in S1, the user writes data using RAID3 mode.

The further technical scheme is as follows: the S3 includes:

s31, the SSD internally and periodically arranges the hot data of part of the first space;

s32, changing the hot data to cold data and triggering a data movement operation.

The further technical scheme is as follows: after S4, the method further includes: and when the user needs to read data subsequently, reading from the second space.

The further technical scheme is as follows: the user read data is read using RAID5 mode.

A system for improving the read-write performance of a solid state disk comprises an acquisition unit, a storage management unit, a sorting triggering unit and a moving storage unit;

the acquisition unit is used for acquiring data newly written by a user;

the storage management unit is used for storing the new data into the first space by the SSD and managing the new data as hot data;

the arrangement triggering unit is used for periodically arranging the hot data in part of the first space in the SSD to change the hot data into cold data and triggering data transfer;

the moving storage unit is used for moving the cold data from the first space to the second space and storing the cold data.

The further technical scheme is as follows: the sorting triggering unit comprises a sorting module and a triggering module;

the sorting module is used for periodically sorting the thermal data of part of the first space in the SSD;

and the trigger module is used for changing the hot data into the cold data and triggering the data moving operation.

The further technical scheme is as follows: the reading unit is used for reading from the second space when a user needs to read data subsequently.

Compared with the prior art, the invention has the beneficial effects that: through separately managing different data in the solid state disk, the first space and the second space use different RAID modes for data protection, the RAID3 which does not influence the write performance is used for the first space for storing hot data, and the RAID5 which does not influence the read performance is used for the second space for storing cold data, so that the maximization of the whole read-write performance is obtained, and the requirement can be better met.

The invention is further described below with reference to the accompanying drawings and specific embodiments.

Drawings

FIG. 1 is a flowchart illustrating a method for improving read/write performance of a solid state disk according to the present invention;

FIG. 2 is a schematic diagram of a data storage space;

FIG. 3 is a schematic diagram of data storage in RAID3 mode;

FIG. 4 is a schematic diagram of data storage in RAID5 mode;

FIG. 5 is a schematic diagram of a user read operation in RAID3 mode;

FIG. 6 is a schematic diagram of multi-plane write stitching under RAID5 mode;

FIG. 7 is a schematic diagram of a multi-plane operation;

FIG. 8 is a schematic diagram of multi-plane write stitching in RAID3 mode;

FIG. 9 is a schematic diagram of a user read operation in RAID5 mode;

FIG. 10 is a schematic diagram of use in a hybrid mode;

fig. 11 is a block diagram of a system for improving the read/write performance of a solid state disk according to the present invention.

10 acquisition unit 20 storage management unit

30 collation trigger unit 31 collation module

32 trigger module 40 moves memory cells

50 read unit

Detailed Description

In order to more fully understand the technical content of the present invention, the technical solution of the present invention will be further described and illustrated with reference to the following specific embodiments, but not limited thereto.

As shown in fig. 1 to fig. 11, in the specific embodiment, as shown in fig. 1 to fig. 10, the present invention discloses a method for improving the read-write performance of a solid state disk, including the following steps:

s1, acquiring the data newly written by the user;

s2, the SSD stores the new data to the first space to be managed as hot data;

s4, the cold data is moved from the first space to the second space and stored.

As shown in fig. 2, wherein the hot data is the data newly written by the user.

The cold data is valid data moved during storage of the data in the SSD.

The first space is mainly used for storing data which is written by a user and is called hot data (hot data); the second space is mainly used for storing valid data moved during data arrangement inside the SSD, and is called cold data (cold data).

As shown in fig. 3 and 4, the present embodiment adopts RAID3 and RAID5, which are widely used, and the distribution thereof is shown in the figure, and the storage distribution of the check data is different in different RAID modes.

In S1, the user writes data in the RAID3 mode without affecting the write performance.

Wherein S3 includes:

Wherein, after S4, the method further comprises: when the user needs to read data subsequently, reading is carried out from the second space, so that the influence of the RAID behavior inside the SSD on the reading performance is minimized.

In this embodiment, the user reads the data in RAID5 mode.

As shown in fig. 5, for the distribution mode of RAID3, the check data monopolizes one parallelism stripe of the bottom-layer NAND flash, and when reading subsequent data, since the check data is not needed when normal data is read at a time when no error occurs, the stripe where the check data is located is always in an idle state, which results in that the advantage of multiple parallelisms of the NAND flash in the SSD cannot be fully embodied, and if the parallelism of the bottom-layer NAND flash is N, the read performance at this time is (N-1)/N times that when the RAID function is not included.

As shown in fig. 6, for the distribution of RAID5, when data is written, since check data needs to be written last in the entire RAID stripe, for the data writing order in a single stripe, D0_0, D0_1, and D0_2 are written in sequence, and the check data is written last; this results in that the data block of stripe D0_0 in the figure cannot implement multi-plane write (multi-plane write: a NAND flash operation behavior that can reduce the time consumption of write operation) operation during NAND flash write, thereby affecting the overall write performance.

As shown in fig. 7, when Multi-plane write operation indicates that data is written at a NAND flash end, if two planes (bottom-layer NAND flash operation channel units) under a certain lun (bottom-layer NAND flash operation channel unit) of a certain die (bottom-layer NAND flash operation channel unit) are simultaneously written, mult i-plane operation provided by the NAND flash may be used, and the operation binds write behaviors under the two planes, so that the execution time of the two write behaviors may be accelerated. It can be seen that RAID3 and RAID5 both have an impact on performance in different read and write scenarios, but have advantages with respect to each other.

As shown in FIG. 8, in the write scenario, the parity data of RAID3 is always distributed at the end of the stripe, which makes the write of parity data not affect the execution of mult i-plane, and in the figure, parity data P0_3 can be written into NAND flash through mult i-plane write together with the last data block D0_2 in the stripe.

As shown in fig. 9, in the read scenario, the check data of RAID5 is uniformly distributed under each plane, which makes there be no idle plane when the user reads, so that the read performance does not have a large impact compared to the non-RAID mode.

Based on the premise, the invention uses different RAID modes on different data management spaces in the SSD, thereby realizing the maximization of the overall performance.

As shown in fig. 10, the RAID3 mode that does not affect the write performance is adopted for the first space storing hot data because user data is mainly written, and the RAID5 that does not affect the read performance is adopted for the second space storing cold data because user data is mainly read; the data writing in the second space is mainly data transfer inside the SSD, and in general, the user has little influence on the writing performance of this portion.

In the SSD, the management operation of the user data is as follows: when a user newly writes data, the SSD stores the new data in a first space to be managed as hot data, the first space is managed by adopting the RAID3 model which does not influence the writing performance, along with the continuous writing of the user data, part of the hot data in the first space is changed into cold data, a data moving operation is triggered, the data moving operation moves the hot data from the first space to a second space, and the second space is managed by adopting the RAID5 model which does not influence the reading performance; when a subsequent user reads data, the data is mainly read from the second space, so that the influence of the RAID behavior inside the SSD on the reading performance is minimized.

As shown in fig. 11, the present invention discloses a system for improving the read-write performance of a solid state disk, which includes an obtaining unit 10, a storage management unit 20, a sorting triggering unit 30, and a moving storage unit 40;

an obtaining unit 10, configured to obtain data newly written by a user;

a storage management unit 20, configured to the SSD to store the new data in the first space, and manage the new data as hot data;

the arrangement triggering unit 30 is used for periodically arranging the hot data in part of the first space in the SSD to change the hot data into cold data and trigger data transfer;

and a transfer storage unit 40 for transferring the cold data from the first space to the second space and storing the cold data.

The sorting triggering unit 30 comprises a sorting module 31 and a triggering module 32;

the sorting module 31 is used for periodically sorting the thermal data of part of the first space in the SSD;

and the triggering module 32 is used for changing the hot data into cold data and triggering data moving operation.

The system for improving the read-write performance of the solid state disk further includes a reading unit 50, configured to read data from the second space when a user needs to read data subsequently.

According to the invention, different data in the solid state disk are separately managed, the first space and the second space use different RAID modes for data protection, the RAID3 which does not influence the write performance is used for the first space for storing hot data, and the RAID5 which does not influence the read performance is used for the second space for storing cold data, so that the maximization of the whole read-write performance is obtained, and the requirement can be better met.

The technical contents of the present invention are further illustrated by the examples only for the convenience of the reader, but the embodiments of the present invention are not limited thereto, and any technical extension or re-creation based on the present invention is protected by the present invention. The protection scope of the invention is subject to the claims.

Claims

1. A method for improving the read-write performance of a solid state disk is characterized by comprising the following steps:

s1, acquiring the data newly written by the user;

s2, the SSD stores the new data to the first space to be managed as hot data;

s4, moving the cold data from the first space to the second space and storing the cold data;

the hot data is newly written data by a user; the cold data is effective data moved during the process of storing the data in the SSD; in the S1, the user writes data by adopting a RAID3 mode; after S4, the method further includes: when the user needs to read data subsequently, reading from the second space; the user read data is read using RAID5 mode.

2. The method according to claim 1, wherein the step S3 includes:

3. A system for improving the read-write performance of a solid state disk is characterized by comprising an acquisition unit, a storage management unit, a sorting triggering unit and a moving storage unit;

the acquisition unit is used for acquiring data newly written by a user;

the moving storage unit is used for moving cold data from the first space to the second space and storing the cold data;

the hot data is newly written data by a user; the cold data is effective data moved during the process of storing the data in the SSD; in the acquisition unit, a user writes data in a RAID3 mode; the data reading device further comprises a reading unit, and the reading unit is used for reading from the second space when a user needs to read data subsequently, and the user read data is read in a RAID5 mode.

4. The system for improving the read-write performance of the solid state disk according to claim 3, wherein the sorting trigger unit comprises a sorting module and a trigger module;