WO2016134571A1

WO2016134571A1 - Raid-based storage method and storage apparatus

Info

Publication number: WO2016134571A1
Application number: PCT/CN2015/082392
Authority: WO
Inventors: 张翼
Original assignee: 中兴通讯股份有限公司
Priority date: 2015-02-25
Filing date: 2015-06-25
Publication date: 2016-09-01
Also published as: CN105988727B; CN105988727A

Abstract

A redundant array of independent disk (RAID)-based storage method and apparatus. The method comprises: dividing a hard disk stored according to traditional RAID into a plurality of storage blocks; combining the storage blocks to obtain a resource pool; and using the resource pool obtained through the combination to establish a corresponding logical unit number (LUN), wherein the LUN and an original LUN in the hard disk stored according to traditional RAID occupy the same hard disk physical space.

Description

RAID-based storage method and storage device

Technical field

This document relates to the field of hard disk storage, and in particular to a RAID-based storage method and storage device.

Background technique

The RAID (Redundant Array of Independent Disk) technology was born in 1987 and was proposed by the University of California, Berkeley. The original purpose of the development was to replace the large expensive hard disk with a small and expensive hard disk to reduce the cost of large-scale data storage. At the same time, it also hopes to use redundant information, so that the access to the data will not be lost when the hard disk fails. In order to develop a certain level of data protection technology, and can appropriately increase the data transmission speed. In RAID technology, by placing the same data on multiple hard disks, input and output operations can be balanced in a balanced manner, improving performance, and increasing the mean time between failures due to the use of multiple hard disks. Due to the redundant storage of data, when any one of the hard disks in the array fails, the data in the failed disk can be reconstructed through the data of other hard disks in the storage array.

Referring to FIG. 1 , which is a schematic diagram of a conventional RAID storage in the related art, as shown in FIG. 1 , in a conventional RAID mode, a hard disk is regarded as a basic unit, and a RAID group is created by using multiple hard disks (at least two), as shown in FIG. 1 . It is shown that four hard disks form a RAID group according to the RAID3+1 policy, and then the volume VOL is created by performing space allocation on the basis of the RAID group, thereby obtaining a user LUN (Logical Unit Number) space according to the VOL. The LUN space is user-oriented and is allocated to the user for use according to the user's data storage requirements.

For the traditional RAID system, the data in the reconstruction process is not guaranteed. If a bad disk occurs before the reconstruction is completed, the data will be lost and cannot be recovered. Therefore, for a storage system, the most important sign of its reliability is that the smaller the RAID reconstruction time, the better, thus reducing the probability of bad disk before the reconstruction is completed. With the development of storage technology, traditional RAID technology has been unable to meet the requirements of using large-capacity hard disks in the system. In the past few years, block virtualization storage has been developed based on traditional RAID technology. The block virtualization storage method integrates the functions of data protection and data distribution across hard disks, and fully meets the storage application requirements in the virtual machine environment.

Please refer to FIG. 2 , which is a schematic diagram of a block virtualization storage mode. As shown in FIG. 2 , in the block virtualization mode, the hard disk is broken into blocks of several sizes, and then these small blocks are aggregated into a large resource pool POOL. The creation of the user LUN is to allocate space in this large POOL.

For the user, the service deployed on the traditional RAID cannot meet the performance and function requirements, and the service needs to be switched to the array capable of supporting the block virtualization function. Compared with the traditional RAID technology, The block virtualization storage mode has excellent performances such as hierarchical storage, deduplication, and fast reconstruction. In order to replace the traditional RAID-to-block virtualized storage mode, most vendors usually perform the replacement process in two ways: one way is to perform hardware upgrades, such as purchasing new hardware supporting the block virtualization function, and then upgrading. To perform data migration between storage arrays, you need to migrate the service data of the original RAID to a new device and then reconfigure it, and finally relocate its services to the new storage array. The other way is, the previous storage. The array itself is an array device that supports both traditional RAID and virtualized POOL functions. Then, the virtualized POOL can be newly created in the original array, and the data on the original traditional RAID can be migrated to the new POOL through the data migration method in the array. This enables storage to migrate to new block virtualization.

Both of the above methods involve a problem: data migration, one is data migration between arrays, and the other is data migration within the array, both of which are reflected in long migration time and system reliability during migration. The risk is reduced because the data disk failure may occur during the migration process, which leads to a decrease in system reliability. At the same time, the above two methods occupy a large amount of system resources during the migration process, thereby affecting system performance.

Summary of the invention

The embodiment of the present invention provides a storage method and a storage device based on a RAID, which are used to solve the problem of long migration time, large system resources, and system reliability caused by data migration in the update of the storage mode in the related art. problem.

The embodiment of the invention provides a RAID-based storage method, including:

Dividing a hard disk according to a conventional RAID storage into a plurality of storage blocks;

Obtaining a resource pool by using the storage block combination;

The LUNs of the hard disk occupy the same physical space as the original LUNs in the traditional RAID storage hard disk.

Optionally, the storage block is a storage block block, and the resource pool is a virtual disk VD space;

The obtaining the resource pool by using the storage block combination includes:

Mapping the block to the LD space according to the mapping relationship between the block and the logical hard disk LD;

The LD space is combined into a VD space according to the RAID algorithm adopted by the hard disk according to the conventional RAID storage.

Optionally, the mapping the block to the LD space includes: adding the LD space to the storage array, and adding the LD space to the hard disk according to the traditional RAID storage using the same RAID policy. To the LD space; when there is no LD space in the storage array and the same RAID policy is established for the hard disk according to the traditional RAID storage, a new LD space is established, and the block is mapped to the LD space. .

Optionally, the method further includes: determining, according to the LUN identifier of the block, the LUN identifier of the small space Chunk in the VD.

Optionally, the determining, according to the LUN identifier of the block, the LUN identifier of the small space Chunk in the VD includes:

According to the LUN identifier of the block, the LUN identifier is used as the LUN identifier of the stripe strip corresponding to the block in the LD space. The LUN identifier is used as the LUN identifier of the Chunk corresponding to the strip in the VD space according to the LUN identifier corresponding to the strip. .

Optionally, the storage block is a logical block CK; and the obtaining the resource pool by using the storage block combination includes:

According to the RAID algorithm adopted by the hard disk according to the traditional RAID storage, logical blocks CK from different hard disks are formed into a logical block group CKG;

The logical block group CKG is divided into a finer-grained logical tile Extent, and the set of the Extent is the resource pool.

Optionally, the method further includes: determining, according to the LUN identifier of the logical block CK, a LUN identifier of the logical small block Extent in the resource pool.

Optionally, the determining, according to the LUN identifier of the logical block CK, the LUN identifier of the logical small block Extent in the resource pool, including:

The LUN identifier is used as the LUN identifier of the logical block group CKG corresponding to the CK according to the LUN identifier of the logical block CK. The LUN identifier is used as the LUN identifier of the Extent corresponding to the CKG in the resource pool according to the LUN identifier of the CKG.

Optionally, the number of LUNs created by using the combined resource pool is the same as the number of original LUNs in the traditional RAID and the physical space occupied by each LUN.

The resource pool obtained by using the combination and the corresponding logical unit number LUN includes: all the resource blocks having the same LUN identifier are allocated to the same LUN according to the LUN identifier of the resource block in the resource pool, where the resource pool is VD In the case of space, the resource block is a Chunk, and when the resource pool is a collection of Extents, the resource block is an Extent.

The embodiment of the present invention further provides a storage device based on an independent redundant disk array RAID, wherein the RAID-based storage device is disposed on a storage array, and the storage device includes:

a storage block generating unit, configured to divide the hard disk stored according to the traditional RAID into a plurality of storage blocks;

a resource pool generating unit configured to obtain a resource pool by using the storage block combination;

The LUN update unit is configured to use the resource pool obtained by the combination to establish a corresponding logical unit number LUN, wherein the LUN occupies the same physical space of the hard disk as the original LUN of the hard disk stored in the traditional RAID.

Optionally, the storage block is a storage block block, and the resource pool is a virtual disk VD space; the resource pool generating unit includes a VD generating unit, and the VD generating unit includes:

The LD generation module is configured to map the block to the LD space according to the mapping relationship between the block and the logical hard disk LD;

The VD generating module is configured to combine the LD spaces into a VD space according to the RAID algorithm adopted by the hard disks stored according to the conventional RAID.

Optionally, the LD generating module is configured to: when an LD space already exists in the storage array, and the LD space is established by using the same RAID policy as the hard disk according to the traditional RAID storage, Adding the block to the LD space; when there is no LD space in the storage array and the same RAID policy is established by using the traditional RAID storage hard disk, a new LD space is established, and the block is mapped. Go to the LD space.

Optionally, the storage device further includes:

The first LUN identifier storage unit is configured to determine the LUN identifier of the small space Chunk in the VD according to the LUN identifier of the block.

Optionally, the first LUN identifier storage unit is configured to: use the LUN identifier as a LUN identifier of a stripe strip corresponding to the block in the LD space according to the LUN identifier of the block; according to the LUN identifier corresponding to the strip, The LUN identifier is used as the LUN identifier of the Chunk corresponding to the Strip in the VD space.

Optionally, the storage block is a logical block CK; the resource pool generating unit includes:

The CKG generating module is configured to form the logical block CKG from the different hard disks according to the RAID algorithm adopted by the hard disk according to the traditional RAID storage; and

The resource pool generating module is configured to divide the logical block group CKG into a finer-grained logical tile Extent, and the set of the Extent is the resource pool.

Optionally, the storage device further includes:

The second LUN identifier storage unit is configured to determine the LUN identifier of the logical small block Extent in the resource pool according to the LUN identifier of the logical block CK.

Optionally, the second LUN identifier storage unit is configured to: set the LUN identifier as a LUN identifier of a logical block group CKG corresponding to the CK according to the LUN identifier of the logic block CK; according to the LUN identifier of the CKG, The LUN identifier is used as the LUN identifier of the Extent corresponding to the CKG in the resource pool.

The LUN update unit is further configured to: allocate resource blocks having the same LUN identifier to the same LUN according to the LUN identifier of the resource block in the resource pool, where the resource block is a Chunk when the resource pool is a VD space. When the resource pool is a collection of Extents, the resource block is an Extent.

The embodiment of the invention further provides a computer readable storage medium storing program instructions, which can be implemented when the program instructions are executed.

In the RAID-based storage method provided by the embodiment of the present invention, the hard disk stored in the traditional RAID storage in the storage array is divided into multiple storage blocks, and the storage blocks are added to the block virtualization based on the Raid algorithm and the block virtualization storage technology. Storage method in the resource pool. Use the newly added storage space in the resource pool to establish the corresponding LUN. The hard disk space occupied by the newly created LUN is still the hard disk space occupied by the LUN in the original RAID. On the basis of the newly created LUN, the actual hard disk space is partitioned by the storage block, and the correspondence between the storage block and the resource block in the resource pool and the newly created LUN is increased, for example, The relationship between the BLOCK and the Strip, between the Strip and the Chunk, and between the Chunk and the LUN. Therefore, according to these newly established correspondences, the service data stored in the hard disk stored in the conventional RAID is not required to be data. In the case of the migration, the storage method of the block virtualization in the entire column is implemented, and the RAID-based storage method provided by the embodiment of the present invention is provided by the hierarchical storage and the fast reconstruction which can be realized by the virtualized storage mode. In the case that data migration is not required, the update of the block virtualization storage mode is realized, and the traditional RAID data can be directly converted into a storage capable of supporting the virtualized functional data, and the conversion process is updated with respect to the traditional data migration. The time taken by the method can be considered to be completed in an instant, thereby improving the entire system in which the storage array is located. Reliability and ease of use.

BRIEF abstract

1 is a schematic diagram of a conventional RAID storage in the related art;

2 is a schematic diagram of the principle of a block virtualization storage method;

FIG. 3 is a schematic flowchart of a RAID-based storage method according to an embodiment of the present disclosure;

4A and 4B are schematic diagrams of a RAID-based storage method according to an embodiment of the present invention, in which a part of a hard disk is stored and updated according to a conventional RAID mode, and all hard disks are stored in a block virtualization manner;

FIG. 5 is a schematic structural diagram of a RAID-based storage device according to an embodiment of the present invention.

Embodiments of the invention

The RAID-based storage method and storage device provided by the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

FIG. 3 is a schematic flowchart of a RAID-based storage method according to an embodiment of the present invention. As shown in FIG. 3, the method includes:

Step S10, dividing the hard disk stored according to the traditional RAID into a plurality of storage blocks;

The traditional RAID storage described in this document refers to the way that the RAID is created according to a single physical disk. The virtualized storage in this document refers to the RAID created by creating a RAID according to a granularity. The granularity refers to a continuous segment in the physical disk. space.

Step S20, using the storage block combination to obtain a resource pool;

In step S30, the resource pool of the combination is used to establish a corresponding logical unit number LUN, wherein the LUN occupies the same hard disk physical space as the original LUN of the hard disk stored in the traditional RAID.

The RAID-based storage method provided by the embodiment of the present invention is described based on the following block virtualization storage technology. The following describes an exemplary block virtualization storage technology, which is a block composed of N hard disks. Virtualization arrays are introduced, including:

(1.1) dividing N hard disks into a plurality of storage small blocks Block;

(1.2) For these storage tile blocks, according to the first mapping relationship (for example, a discrete algorithm), mapping to the LD space, wherein the smallest unit of the LD space is a stripe strip. In this process, the mapping between Block and Strip will be established.

(1.3) Based on the LD space, according to the RAID policy, the virtual disk VD space is combined, wherein the smallest unit in the VD space is a small space Chunk, and the collection of all Chunk is a resource pool POOL.

The VD space can be obtained according to different RAID policies. For example, the first VD space and the second VD space are respectively established based on the LD space, where the first VD space is obtained according to the RAID 3+1 policy combination. The second VD space is obtained by combining the RAID 2+1 policies.

For convenience of description, the LD space in which the first VD space is combined is referred to as a first LD space, and the LD space in which the second VD space is combined is referred to as a second LD space, that is, the first LD space is classified as RAID 3 The +1 strategy is combined into a first VD space, and the second LD space is combined into a second VD space according to a RAID 2+1 policy.

For the entire POOL, some of the Chunk are based on the RAID 3+1 strategy, and some Chunk are based on the RAID 2+1 strategy.

Different RAID policies correspond to different RAID algorithms. After the RAID policy is determined, the correspondence between the Chunk and the Strip is calculated according to the corresponding RAID algorithm.

(1.4) When the LUN is established, the Chunk in the POOL is assigned to each LUN according to the second mapping relationship (for example, a discrete algorithm). In this process, the mapping relationship between the Chunk and the LUN is established.

The storage array software does not distinguish between different VD spaces when a LUN is created according to the VD space. That is, a Chunk according to different RAID policies may be allocated to the same LUN.

In the embodiment of the present invention, there are two types of LUNs. When a LUN is created, the storage array software allocates a physical space for the LUN. The other LUNs cannot occupy the physical space. The storage array software allocates a small amount of physical space to the LUN, and then allocates real physical space to the LUN as the actual data storage needs. In the second case, the LUN can be created according to the actual physical space occupied by the original LUN of the conventional RAID, and the LUN is created according to the VD and the original LUN. The actual physical address is the same.

Based on the foregoing block virtualization storage technology, a RAID-based storage method provided by an embodiment of the present invention includes:

Step S110, dividing the hard disk according to the traditional RAID storage into a plurality of storage blocks Block;

Step S120, using the block combination to obtain a corresponding virtual hard disk VD space based on the Raid algorithm and the block virtualization storage technology;

The above steps S110 and S120 correspond to step S10 and step S20, respectively.

Wherein, step S120 includes:

Step S1210: Map the Block to the LD space according to the mapping relationship between the Block and the logical hard disk LD.

The basic unit in the LD space is Strip. In the above process, the mapping relationship between Block and Strip will be established.

The mapping of the block to the LD space includes: case 1: when the LD space already exists in the storage array, and the LD space is established by using the same RAID policy as the hard disk stored according to the traditional RAID, adding the block to the The LD space; Case 2: when there is no LD space in the storage array and the same RAID policy is established by using the traditional RAID storage hard disk, a new LD space is established, and the block is mapped to the LD space.

The RAID policy in the embodiment of the present invention refers to establishing a RAID storage array by using different RAID algorithms. For example, the RAID policy may be a RAID 3+1 policy or a RAID 2+1 policy. A RAID policy can also refer to the establishment of RAID storage arrays using different RAID standards, such as RAID6, RAID10, and so on.

In step S1220, the LD space is combined into a virtual hard disk VD space according to a RAID algorithm. The RAID algorithm is the same as the RAID algorithm used by the hard disk according to the conventional RAID storage, that is, the LD space is used. The RAID policy adopted when combining VD space is the same as the RAID policy adopted by the hard disk according to the conventional RAID storage.

The VD space is the resource pool POOL. When a LUN is created according to user requirements, the Chunk is allocated to the corresponding LUN in the smallest unit.

In the process of mapping the block to the LD space, if the situation is one, the space size of the new LD relative to the original LD will increase, and the space size of the new VD corresponding to the new LD will also increase, that is, the space address will be In the new VD, the newly added LD creates a newly added Chunk according to the RAID algorithm, and adds the newly added Chunk to the resource pool POOL; if it is case 2, the newly created LD corresponds to the Chunk to be added to Resource pool POOL.

The Chunk in the resource pool POOL may be generated based on different Raids. According to the system settings, Chunks established according to different RAID policies may be assigned to the same LUN.

Since there is a corresponding relationship between the Block and the Strip, and between the Strip and the Chunk, there is a corresponding relationship between the Block and the Chunk.

Step S130, using the newly added VD in the resource pool to establish a corresponding LUN. The LUN occupies the same physical space of the hard disk as the original LUN in the hard disk stored in the traditional RAID.

Step S130 corresponds to step S30.

In this step, the storage array software establishes a mapping table between the VD and the LUN, that is, establishes the correspondence between the Chunk and the LUN in the VD.

The number of LUNs established by using the newly added VD is the same as the number of LUNs established in the hard disk of the conventional RAID. In the hard disk, the storage space that belongs to the LUN is stored in the virtualized storage mode. It still belongs to the newly created corresponding LUN. For example, if two LUNs are set in the traditional RAID storage, the size is 10G and 20G respectively. In the block virtualization mode, the number of LUNs created by using the newly added VD in the resource pool is also two. The size of the hard disk space occupied by the two newly created LUNs is still the hard disk space occupied by the two LUNs in the original RAID storage. However, for the newly created LUN, the actual hard disk space is divided by the BLOCK method, and the correspondence between the BLOCK and the Strip, between the Strip and the Chunk, and between the Chunk and the LUN is increased. According to these newly established correspondences, the service data stored in the hard disk stored in the traditional RAID storage can realize the storage mode of the block virtualization in the entire column without requiring data migration, thereby providing virtualized storage. Features such as hierarchical storage and fast reconstruction that can be achieved by the method.

To ensure that the newly created LUNs are the same as the hard disk space occupied by the original LUNs in the traditional RAID, you need to determine the LUN ID of the Chunk in the VD based on the LUN ID of the block.

According to the LUN identifier of the block, the LUN identifier is used as the LUN identifier of the strip corresponding to the block in the LD space. The LUN identifier is used as the LUN identifier of the Chunk corresponding to the strip in the VD space according to the LUN identifier corresponding to the strip.

After the LUN ID of the Chunk is determined in the VD space, the Chunk with the same LUN ID is assigned to the same LUN according to the LUN ID of the Chunk in the VD. In this way, the newly created LUN is the same as the hard disk space occupied by the original LUN in the traditional RAID storage hard disk, which avoids the migration of service data and completes the corresponding block virtualization mode. The establishment of the relationship.

After the LUN corresponding to the VD is created, the original LUN corresponding to the hard disk of the traditional RAID is switched to the LUN corresponding to the VD.

Through the above process, the hard disk according to the traditional RAID storage can be converted into the storage mode according to the block virtualization, so that the original service data has the functions of hierarchical storage and rapid reconstruction without data migration.

The function of the tiered storage means that the storage software migrates data with high access frequency, that is, hotspot data, to a hard disk space with better read/write performance according to the frequency of access of the service data in the hard disk. For example, the hard disk in the entire column can be stored. It includes one or more of SSD (High Performance Layer), SAS (Performance Layer) or NL-SAS (Capacity Layer), in which SSD has better read and write performance. When the storage software detects that the service data in any VD is hotspot data, and the hard disk corresponding to the VD is not an SSD hard disk, the data on the VD may be migrated to the idle VD corresponding to the SSD hard disk. Therefore, by converting the hard disk according to the traditional RAID storage into the storage mode according to the block virtualization, the original stored service data is in a new storage mode, and the hotspot data can be migrated to other hard disks, thereby making the original service. The data has the function of tiered storage.

In addition, the storage method of block virtualization also has the advantage of rapid reconfiguration. In the hard disk according to the traditional RAID storage, if one of the hard disks fails, the fault disk needs to be replaced, and data reconstruction is performed in the process of data reconstruction. The data in the newly replaced hard disk needs to be rewritten according to the data in the other hard disks in the RAID group. The RAID group refers to a group of hard disks stored according to the RAID, for example, according to the RAID 3+1 policy. A block hard disk can be called a RAID group. In the process of data reconstruction, data needs to be read from three hard disks and data is written to one disk. After the block is virtualized, the RAID groups are actually distributed. In the multiple hard disks of the entire resource pool POOL, ideally, data can be read from the hard disk involved in the entire resource pool, and data is written into multiple hard disks, so the number of hard disks participating in reading and writing is increased, so , improve the speed of data reconstruction.

The RAID-based storage method in the embodiment of the present invention may be applied to, but not limited to, the following storage arrays: some hard disks in the storage array are stored according to a conventional RAID mode, and some hard disks are stored in a block virtualization manner, and are in the storage array. Some areas of the storage have been stored with business data; all the hard disks in the storage array are stored in the traditional RAID mode, and some of the storage arrays are empty. Business data is already stored.

It should be noted that, if it is another block virtualization storage technology, the storage mode may be converted by using the RAID-based storage method provided by the embodiment of the present invention. The RAID-based storage method provided by the embodiment of the present invention is described below in conjunction with another block virtualization mode.

The following describes an exemplary block virtualization storage technology in the embodiment of the present invention, and introduces a block virtualization array composed of N hard disks, including:

(2.1) dividing the N hard disks into a plurality of logic blocks CK; for example, the size of the logic block CK may be 64M;

(2.2) logical blocks CK from different hard disks are grouped into logical block groups CKG in a RAID mode (such as RAID5, RAID6);

(2.3) The logical block group CKG is divided into a fine-grained logical small block Extent, and the set of the Extent is a resource pool. For example, the size of the logical small block Extent may be 512K to 64MB, for example, it may be configured as 4MB as needed.

The logical small block Extent constitutes the resource pool POOL, and the logical small block Extent is the basic unit when the user space is allocated. When the LUN is created according to the user needs, the logical small block Extent in the resource pool can be allocated to the corresponding LUN.

In step S210, the hard disk stored according to the conventional RAID is divided into a plurality of logical blocks CK.

Step S220, using the Raid algorithm and the block virtualization storage technology, using the resource pool composed of the Extent obtained by the CK combination;

The above steps S210 and S220 correspond to step S10 and step S20, respectively.

Step S220 includes:

Step S2210: The logical blocks CK from different hard disks are formed into a logical block group CKG according to the RAID algorithm adopted by the hard disks stored in the conventional RAID. In the above process, the correspondence between Ck and CKG will be established.

Step S2220, cutting the logical block group CKG into a finer-grained logical block Extent, The set of Extent is the resource pool.

When a LUN is created based on the user's requirements, the Extent in the resource pool is allocated to the corresponding LUN.

Since there is a corresponding relationship between CK and CKG, and between CKG and Extent, there is a corresponding relationship between CK and Extent.

The obtaining the resource pool by using the storage block combination further includes: determining, according to the LUN identifier of the logical block CK, the LUN identifier of the logical small block Extent in the resource pool;

In step S230 (corresponding to step S30), the corresponding LUN is established by using the newly added Extent in the resource pool.

The following describes two application scenarios.

scene one

In this scenario, some of the hard disks in the storage array are stored in the traditional RAID mode. Some of the hard disks are stored in a virtualized manner. According to the service requirements, all the disks in the storage array need to be stored in a virtualized manner.

Referring to FIG. 4A and FIG. 4B, in a RAID-based storage method according to an embodiment of the present invention, a portion of a hard disk is stored and updated according to a conventional RAID mode, and the entire hard disk is stored in a block virtualization manner, as shown in FIG. 4A. The original part of the storage array is based on the traditional Raid part based on the block virtualization. The storage array includes 9 hard disks and D is the hard disk label. The 4 hard disks D1 to D4 are stored in the traditional RAID mode. D5～ The five hard disks of D9 are stored in block virtualization mode; as shown in the external part of the dotted line frame, a block virtualized array consisting of five hard disks, wherein (1) D5 to D9 are divided into five hard disks. Blocks; (2) for these blocks, according to the first mapping relationship (discrete algorithm), mapped to the LD space, (3) based on the LD space, respectively establish a first VD space and a second VD space, wherein the second VD The space is obtained by combining the RAID 3+1 policies. The first VD space is obtained according to the RAID 2+1 policy combination. For the convenience of description, the LD space that combines the first VD space is called the first LD space. Between the four columns on the left side of the figure, the LD space that combines to obtain the second VD space is called the second LD space, that is, the three columns on the right side of the figure, that is, the first LD space is in accordance with the RAID 3+1 strategy. The combination is a first VD space, and the second LD space is combined into a second VD space according to a RAID 2+1 policy. (4) The first VD space and the second VD space are allocated to each LUN according to the second mapping relationship, respectively. The second mapping relationship may also be a discrete algorithm.

It should be noted that, before applying the RAID-based storage method provided by the embodiment of the present invention, the storage array has been set up in the above manner. And some of the four hard disks stored in the traditional RAID mode have been used to store business data.

The RAID-based storage method of the embodiment of the present invention is described below in conjunction with the storage array in the above scenario. As shown in FIG. 4B, the storage array is updated to be stored in a block virtualization manner, wherein:

Step S11: The hard disk stored in the traditional RAID mode is divided into multiple blocks. That is, four hard disks stored based on the traditional RAID mode are divided into multiple storage blocks;

In step S12, the block is added to the corresponding LD space according to the first mapping relationship between the block and the logical hard disk LD. In this embodiment, four hard disks are stored based on the RAID 3+1 policy, and the first LD space is The LD space corresponding to the storage block is the first LD space, that is, the block corresponding to the block is added to the first LD space.

Step S13: Combine the new first LD space into a new first VD space according to the RAID algorithm; as shown in FIG. 4B, the new VD space becomes larger than the original VD space, wherein the outside of the dotted line The VD space is the original VD space, and the VD space in the dotted frame is the newly added VD space;

Step S14: Mapping the new first VD space to the LUN according to the mapping relationship between the VD space and the storage volume LUN;

The new LUN is replaced with the corresponding LUN in the traditional Raid. The size of the new LUN is unchanged from the original LUN, and the new LUN is actually pointed to the original LUN. The same hard disk space, but the new LUN is built on the basis of block virtualization, so it has hierarchical storage and fast reconfiguration. The previous business data is still stored in In the new LUN, as the service is updated and expanded, data on the new LUN is continuously migrated to achieve data dispersion throughout the pool, thereby improving the reliability requirements and reconstruction speed of the entire system.

Scene two

In this embodiment, all the hard disks in the storage array are stored in the traditional RAID mode. According to the service requirements, all the hard disks in the storage array need to be stored in a virtualized manner.

The storage array includes four hard disks, and four hard disks are stored in the traditional RAID mode. Among them, four hard disks are stored based on the RAID 3+1 policy. The RAID-based storage method of the embodiment of the present invention is described below in conjunction with the foregoing scenarios. among them:

Step S21, dividing four hard disks stored in the traditional RAID mode into multiple blocks;

Step S22: Add a Block to the LD space according to a first mapping relationship between the Block and the logical hard disk LD.

Step S23, combining the foregoing LD spaces into a VD space according to a RAID algorithm;

Step S24: Map the VD space to each LUN according to the mapping relationship between the VD space and the storage volume LUN.

An embodiment of the present invention further provides a RAID-based storage device, where the RAID-based storage device is disposed on a storage array. FIG. 5 is a schematic structural diagram of a RAID-based storage device according to an embodiment of the present invention. As shown in FIG. 5, the storage device includes:

The storage block generating unit 10 is configured to divide the hard disk stored according to the conventional RAID into a plurality of storage blocks.

a resource pool generating unit 20, configured to obtain a resource pool by using the storage block combination based on an algorithm Raid and a block virtualization storage manner;

The LUN update unit 30 is configured to use the resource pool obtained by the combination to establish a corresponding logical unit number LUN, wherein the LUN occupies the same hard disk physical space as the original LUN in the traditional RAID storage hard disk.

Optionally, the storage block is a storage block block, and the resource pool is a VD space; the resource pool generating unit 20 includes a VD generating unit, and the method includes:

The LD generating module 21 is configured to map the block to the LD space according to the mapping relationship between the block and the logical hard disk LD;

The VD generating module 22 is configured to combine the LD space into a virtual hard disk VD space according to a RAID algorithm; wherein the RAID algorithm is the same as the RAID algorithm adopted by the hard disk according to the conventional RAID storage.

Optionally, the LD generating module 21 is configured to: when the LD space already exists in the storage array, and the LD space is established by using the same RAID policy as the hard disk stored according to the traditional RAID, add the block to the LD space; when there is no LD space in the storage array and the same RAID policy is established by the hard disk of the traditional RAID storage, a new LD space is established, and the block is mapped to the LD space.

Optionally, in the embodiment of the present invention, the storage device further includes:

The first LUN identifier storage unit 40 is configured to determine a LUN identifier of the Chunk in the VD according to the LUN identifier of the block;

The LUN identifier storage unit 40 is configured to: use the LUN identifier as the LUN identifier of the strip corresponding to the block in the LD space according to the LUN identifier of the block; and use the LUN according to the LUN identifier corresponding to the strip. Identifies the LUN ID of the Chunk that corresponds to the Strip in the VD space. In the embodiment of the present invention, the number of LUNs established by using the VDs obtained by the combination is the same as the number of LUNs established by the hard disks stored in the conventional RAID. The LUN update unit 30 is further configured to: according to the LUN identifier of the Chunk in the VD. , the Chunk with the same LUN ID is assigned to the same LUN.

In the embodiment of the present invention, when the storage block is a logical block CK, the resource pool generating unit 20 includes:

The CKG generating module 26 is configured to form the logical block CKG from the different hard disks according to the RAID algorithm adopted by the hard disk according to the traditional RAID storage; and

The resource pool generating module 27 is configured to divide the logical block group CKG into a finer-grained logical tile Extent, and the set of the Extent is the resource pool.

Optionally, the storage device further includes: a second LUN identifier storage unit, configured to determine a LUN identifier of the logical small block Extent in the resource pool according to the LUN identifier of the logical block CK;

The second LUN identifier storage unit is configured to: set the LUN identifier as a LUN identifier of the logical block group CKG corresponding to the CK according to the LUN identifier of the logical block CK; and use the LUN according to the LUN identifier of the CKG Identifies the LUN ID of the Extent that corresponds to the CKG in the resource pool.

In the embodiment of the present invention, the LUNs created by using the combined resource pool are the same as the number of original LUNs in the hard disk and the physical space occupied by each LUN. The LUN update unit 30 is also configured to All the resource blocks with the same LUN identifier are allocated to the same LUN according to the LUN identifier of the resource block in the resource pool. When the resource pool is a VD space, the resource block is a Chunk, and when the resource pool is an Extent set. The resource block is an Extent.

One of ordinary skill in the art will appreciate that all or a portion of the above steps may be accomplished by a program that instructs the associated hardware, such as a read-only memory, a magnetic disk, or an optical disk. Alternatively, all or part of the steps of the above embodiments may also be implemented using one or more integrated circuits. Correspondingly, each module/unit in the foregoing embodiment may be implemented in the form of hardware or in the form of a software function module. Embodiments of the invention are not limited to any specific form of combination of hardware and software.

Industrial applicability

The RAID-based storage method provided by the embodiment of the present invention implements the update of the block virtualized storage mode without data migration, and realizes that the traditional RAID data can be directly converted into a storage capable of supporting the virtualized functional data. This improves the reliability and ease of use of the entire system in which the storage array resides.

Claims

A storage method based on an independent redundant disk array RAID, the method comprising:

Dividing a hard disk according to a conventional RAID storage into a plurality of storage blocks;

Obtaining a resource pool by using the storage block combination;

The LUNs of the hard disk occupy the same physical space as the original LUNs in the traditional RAID storage hard disk.
The method of claim 1, wherein the memory block is a memory block block, and the resource pool is a virtual disk VD space;

The obtaining the resource pool by using the storage block combination includes:

Mapping the block to the LD space according to the mapping relationship between the block and the logical hard disk LD;

The LD space is combined into a VD space according to the RAID algorithm adopted by the hard disk according to the conventional RAID storage.
The method of claim 2, wherein the mapping the block to the LD space comprises: when an LD space already exists in the storage array, and the LD space is established with the same RAID policy as the hard disk according to the conventional RAID storage Adding the block to the LD space; when there is no LD space in the storage array and the same RAID policy is established by using the traditional RAID storage hard disk, a new LD space is established, and the Block is mapped to the LD space.
The method of claim 3, further comprising: determining a LUN identifier of the small space Chunk in the VD according to the LUN identifier of the block.
The method of claim 4, wherein determining the LUN identifier of the small space Chunk in the VD according to the LUN identifier of the block comprises:

According to the LUN identifier of the block, the LUN identifier is used as the LUN identifier of the stripe strip corresponding to the block in the LD space. The LUN identifier is used as the LUN identifier of the Chunk corresponding to the strip in the VD space according to the LUN identifier corresponding to the strip. .
The method of claim 1, wherein the storage block is a logical block CK; and the utilizing the storage block combination to obtain a resource pool comprises:

According to the RAID algorithm adopted by the hard disk according to the traditional RAID storage, logical blocks CK from different hard disks are formed into a logical block group CKG;

The logical block group CKG is divided into a finer-grained logical tile Extent, and the set of the Extent is the resource pool.
The method according to claim 6, further comprising: determining, according to the LUN identifier of the logical block CK, a LUN identifier of the logical small block Extent in the resource pool.
The method of claim 7, the determining, according to the LUN identifier of the logical block CK, the LUN identifier of the logical small block Extent in the resource pool, comprising: corresponding to the LUN according to the LUN identifier of the logical block CK The LUN ID of the CKG of the logical block group; the LUN identifier of the CKG is used as the LUN identifier of the Extent corresponding to the CKG in the resource pool.
The method of claim 4 or 5 or 7 or 8, wherein the number of LUNs created by using the combined resource pool and the number of original LUNs in the hard disk stored in the conventional RAID and the physical space occupied by each LUN the same;

The resource pool obtained by using the combination and the corresponding logical unit number LUN includes: all the resource blocks having the same LUN identifier are allocated to the same LUN according to the LUN identifier of the resource block in the resource pool, where the resource pool is VD In the case of space, the resource block is a Chunk, and when the resource pool is a collection of Extents, the resource block is an Extent.
A storage device based on an independent redundant disk array RAID, the RAID-based storage device being disposed on a storage array, the storage device comprising:

a storage block generating unit, configured to divide the hard disk stored according to the traditional RAID into a plurality of storage blocks;

a resource pool generating unit configured to obtain a resource pool by using the storage block combination;

The LUN update unit is configured to use the resource pool obtained by the combination to establish a corresponding logical unit number LUN, wherein the LUN occupies the same physical space of the hard disk as the original LUN of the hard disk stored in the traditional RAID.
The storage device according to claim 10, wherein said storage block is a storage chunk Block, said resource pool is a virtual disk VD space; said resource pool generating unit comprises a VD generating unit, The VD generating unit includes:

The LD generation module is configured to map the block to the LD space according to the mapping relationship between the block and the logical hard disk LD;

The VD generating module is configured to combine the LD spaces into a VD space according to the RAID algorithm adopted by the hard disks stored according to the conventional RAID.
The storage device of claim 11, wherein the LD generation module is configured to: when an LD space already exists in the storage array, and the LD space is established by using the same RAID policy as the hard disk stored according to the conventional RAID, Adding a block to the LD space; when there is no LD space in the storage array and the same RAID policy is established for the hard disk according to the traditional RAID storage, a new LD space is established, and the block is mapped to the LD space. Said LD space.
The storage device of claim 12, the storage device further comprising:

The first LUN identifier storage unit is configured to determine the LUN identifier of the small space Chunk in the VD according to the LUN identifier of the block.
The storage device of claim 13, wherein the first LUN identifier storage unit is configured to: use the LUN identifier as a LUN identifier of a stripe strip corresponding to the block in the LD space according to the LUN identifier of the block; The LUN identifier is used as the LUN identifier of the Chunk corresponding to the Strip in the VD space.
The storage device of claim 10, wherein the storage block is a logical block CK; the resource pool generating unit comprises:

The CKG generating module is configured to form the logical block CKG from the different hard disks according to the RAID algorithm adopted by the hard disk according to the traditional RAID storage; and

The resource pool generating module is configured to divide the logical block group CKG into a finer-grained logical tile Extent, and the set of the Extent is the resource pool.
The storage device of claim 15, the storage device further comprising:

The second LUN identifier storage unit is configured to determine the LUN identifier of the logical small block Extent in the resource pool according to the LUN identifier of the logical block CK.
The storage device of claim 16, wherein the second LUN identifier storage unit is configured to: use the LUN identifier as a logic corresponding to CK according to a LUN identifier of the logic block CK The LUN ID of the CKG of the block group. The LUN ID of the CKG is used as the LUN identifier of the Extent corresponding to the CKG in the resource pool.
The storage device of claim 13 or 14 or 16 or 17, wherein the number of LUNs created by the combined resource pool and the number of original LUNs in the hard disk stored in the conventional RAID and the physics occupied by each LUN The same space;

The LUN update unit is further configured to: allocate resource blocks having the same LUN identifier to the same LUN according to the LUN identifier of the resource block in the resource pool, where the resource block is a Chunk when the resource pool is a VD space. When the resource pool is a collection of Extents, the resource block is an Extent.
A computer readable storage medium storing program instructions that, when executed, implement the method of any of claims 1-9.