WO2019062856A1

WO2019062856A1 - Data reconstruction method and apparatus, and data storage system

Info

Publication number: WO2019062856A1
Application number: PCT/CN2018/108342
Authority: WO
Inventors: 林鹏; 汪渭春; 林起芊
Original assignee: 杭州海康威视系统技术有限公司
Priority date: 2017-09-29
Filing date: 2018-09-28
Publication date: 2019-04-04
Also published as: CN109582213B; CN109582213A

Abstract

A data reconstruction method, apparatus, and system, relating to the technical field of data storage. The method comprises: when detecting the occurrence of a faulty disk storing m data blocks in a first storage node, respectively sending a reconstruction command to n target storage nodes amongst a plurality of storage nodes, the first storage node being any storage node amongst the plurality of storage nodes, and the n reconstruction commands sent to the n target reconstruction nodes being used for instructing: implementing reconstruction of m data blocks and storing same, different storage nodes to the first storage node being present amongst the n target storage nodes, m≥n≥1 (301); and sending an acquisition command to the first storage node, the acquisition command being used for instructing the first storage node to acquire and store the m data blocks reconstructed by the n target storage nodes (302). The present method solves the problem of low data reconstruction efficiency in data storage systems and increases the efficiency of data reconstruction in data storage systems; the method is used for data reconstruction.

Description

Data reconstruction method and device, data storage system

Cross-reference to related applications

The present disclosure claims the priority of the Chinese Patent Application No. 201710903893.4 filed on Sep. 29, 2017, the entire disclosure of which is hereby incorporated by reference.

Technical field

The present disclosure relates to the field of data storage technologies, and in particular, to a data reconstruction method and apparatus, and a data storage system.

Background technique

With the development of data storage technology, the data storage system based on the Serial Attached Small Computer System Interface (SAS) protocol has been widely used.

In the related art, the data storage system based on the SAS protocol includes: a metadata management server (English: Metadata Server; MDS for short), a SAS switch, and a plurality of storage nodes, and the plurality of storage nodes are connected to each other through a SAS switch. Each storage node includes multiple disks. When the user terminal needs to store the target data in the data storage system, a storage node in the data storage system cuts the target data into a plurality of data blocks (also referred to as striping processing on the target data, the plurality of The data blocks belong to the same stripe) and the multiple data blocks are stored on separate disks. The MDS can be used to store stripe information (a kind of metadata) of each data block, and the stripe information includes: a stripe identifier of the data block (that is, an identifier of a strip to which the data block belongs), and a disk identifier where the data block is located. The identifier of the data block, the data volume of the data block, and the erasure code of the data block (English: Erasure Coding; abbreviation: EC) type. When a faulty disk occurs on the storage node, the MDS may send a reconstruction instruction to the storage node, where the reconstruction instruction includes stripe information of each data block stored on the faulty disk, and the storage node may be in accordance with the reconstruction instruction. The stripe information of each data block reconstructs each data block stored on the failed disk, that is, restores each data block.

However, since a large number of data blocks are usually stored on the disk, and when the data processing capability of the storage node where the failed disk is located is weak, the storage node reconfigures the data stored on the failed disk to be slow. The data reconstruction system has low data reconstruction efficiency.

Summary of the invention

The present disclosure provides a data reconstruction method and device, and a data storage system, which can solve the problem of low data reconstruction efficiency of the data storage system. The technical solution is as follows:

In a first aspect, a data reconstruction method is provided for a metadata management server MDS in a data storage system, the data storage system further comprising: a serial connection small computer system interface SAS switch and a plurality of storage nodes, The plurality of storage nodes are connected to each other by the SAS switch, and the method includes:

When detecting the faulty disk in which the m data blocks are stored in the first storage node, respectively sending a reconstruction instruction to the n target storage nodes of the plurality of storage nodes, where the first storage node is ???said one of the plurality of storage nodes, and the n reconstruction instructions sent to the n target reconstruction nodes are used to indicate that the m data blocks are reconstructed and stored, the n targets There is a storage node different from the first storage node in the storage node, m≥n≥1;

Sending an acquisition instruction to the first storage node, where the obtaining instruction is used to instruct the first storage node to acquire and store the m data blocks reconstructed by the n target storage nodes.

Optionally, before the sending the reconfiguration instruction to the n target storage nodes of the plurality of storage nodes, the method further includes:

Determining, as the n target storage nodes, n storage nodes having a smaller load among the plurality of storage nodes;

Or determining, by the n storage nodes preset in the plurality of storage nodes, the n target storage nodes;

Alternatively, the n storage nodes that are closest to the routing distance of the MDS among the plurality of storage nodes are determined as the n target storage nodes.

Optionally, sending the reconfiguration instructions to the n target storage nodes of the multiple storage nodes, respectively, including:

Determining a load of each of the target storage nodes;

Determining, according to the load of the n target storage nodes, at least one data block corresponding to each target storage node, wherein a sum of data amounts of all data blocks corresponding to the target storage node is negatively correlated with a load of the target storage node;

Reconstructing the reconstruction instruction corresponding to each target storage node, where the reconstruction instruction corresponding to each target storage node is used to indicate that the data block corresponding to each target storage node is reconstructed and stored;

Sending its corresponding reconstruction instruction to each of the target storage nodes.

Optionally, after the resizing instructions are sent to the n target storage nodes of the plurality of storage nodes, the method further includes:

Receiving a storage application message sent by the first storage node, where the storage application message includes a total data amount of the m data blocks;

And sending, by the storage request message, a storage instruction to the first storage node, where the storage instruction is used to indicate that the reconstructed m data blocks are stored on a target disk, where the target disk is the first storage node A disk in which the available storage capacity is greater than or equal to the total amount of data.

Optionally, the related data block is not stored on the target disk, and the related data block belongs to the same strip as any one of the m data blocks.

Optionally, after the sending the storage instruction to the first storage node, the method further includes:

Receiving, by the first storage node, the storage information of each of the data blocks, where the storage information of each data block includes: an identifier of the target disk and an identifier of each of the data blocks;

Determining stripe information of each of the data blocks according to the identifier of each of the data blocks in the storage information of each data block;

Modifying, in the stripe information of each data block, an identifier of a disk where each of the data blocks is located, to an identifier of the target disk.

Optionally, each of the plurality of storage nodes includes: a storage disk and a cache disk, each storage node having read permission of the storage disk, and read and write permissions of the cache disk, The n target storage nodes include a cache storage node, and the reconstruction instruction sent to the cache storage node is used to indicate: storing the reconstructed data block in a cache disk of the cache storage node, to the first storage After the node sends the acquisition instruction, the method further includes:

And receiving the obtained completion message sent by the first storage node, where the obtained information is used to indicate that the first storage node has acquired and stored the reconstructed m data blocks;

And sending a delete instruction to the cache storage node, where the delete instruction is used to instruct the cache storage node to delete a data block stored on a cache disk thereof.

In a second aspect, a data reconstruction apparatus is provided for a metadata management server MDS in a data storage system, the data storage system further comprising: a serial connection small computer system interface SAS switch and a plurality of storage nodes, The plurality of storage nodes are connected to each other by the SAS switch, and the method includes:

a first sending module, configured to send a reconfiguration instruction to each of the plurality of storage nodes when detecting a faulty disk in which the m data blocks are stored in the first storage node, where The first storage node is any one of the plurality of storage nodes, and the n reconstruction instructions sent to the n target reconstruction nodes are used to indicate that the m data blocks are reconstructed and Storage, a storage node different from the first storage node exists in the n target storage nodes, m≥n≥1;

a second sending module, configured to send an acquisition instruction to the first storage node, where the obtaining instruction is used to instruct the first storage node to acquire and store the m data blocks reconstructed by the n target storage nodes .

Optionally, m≥n≥2, the data reconstruction device further includes:

a first determining module, configured to determine n storage nodes that are less loaded among the plurality of storage nodes as the n target storage nodes;

Or the second determining module is configured to determine, as the n target storage nodes, the n storage nodes preset in the multiple storage nodes;

Alternatively, the third determining module is configured to determine, as the n target storage nodes, n storage nodes that are closest to the routing distance of the MDS among the plurality of storage nodes.

Optionally, the first sending module is configured to:

Determining a load of each of the target storage nodes;

Optionally, the data reconstruction device further includes:

a first receiving module, configured to receive a storage application message sent by the first storage node, where the storage application message includes a total data volume of the m data blocks;

a third sending module, configured to send a storage instruction to the first storage node according to the storage request message, where the storage instruction is used to indicate that the reconstructed m data blocks are stored on a target disk, the target disk A disk that has a storage capacity greater than or equal to the total amount of data in the first storage node.

Optionally, the data reconstruction device further includes:

a second receiving module, configured to receive storage information of each of the data blocks sent by the first storage node, where the storage information of each data block includes: an identifier of the target disk, and each of the data blocks Identification

a fourth determining module, configured to determine stripe information of each of the data blocks according to the identifier of each of the data blocks in the storage information of each data block;

And a modification module, configured to modify an identifier of the disk where each of the data blocks in the strip information of each data block is located to an identifier of the target disk.

Optionally, each of the plurality of storage nodes includes: a storage disk and a cache disk, each storage node having read permission of the storage disk, and read and write permissions of the cache disk, The n target storage nodes include a cache storage node, and the reconstruction instruction sent to the cache storage node is used to indicate that the reconstructed data block is stored in a cache disk of the cache storage node, and the data reconstruction device further include:

a third receiving module, configured to receive the acquired information message sent by the first storage node, where the obtained information message is used to indicate that the first storage node has acquired and stored the reconstructed m data blocks;

And a fourth sending module, configured to send a delete instruction to the cache storage node, where the delete instruction is used to instruct the cache storage node to delete a data block stored on a cache disk.

In a third aspect, a data storage system is provided, the data storage system comprising: a metadata management server MDS, a plurality of storage nodes, and a serial connection small computer system interface SAS switch, wherein the plurality of storage nodes pass the SAS The switches are interconnected, and the MDS comprises the data reconstruction device of the second aspect.

A fourth aspect provides a computer device including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete communication with each other through a bus; the memory is configured to store the computer program; For executing the program stored on the memory, implementing the method steps described in the first aspect.

In a fifth aspect, a computer readable storage medium is provided having stored therein a computer program, the computer program being executed by a processor to implement the method steps of the first aspect.

In a sixth aspect, a data storage system is provided, the data storage system comprising: a metadata management server MDS, a plurality of storage nodes, and a serial connection small computer system interface SAS switch, wherein the plurality of storage nodes pass the SAS The switches are interconnected and the MDS comprises the computer device of claim 16.

In a seventh aspect, a computer program product is provided, which, when run on a computer, causes the computer to perform the method steps of the first aspect.

The beneficial effects brought by the technical solutions provided by the present disclosure are:

When detecting the faulty disk in the first storage node, the MDS sends a reconfiguration instruction to the n target storage nodes, so that the n target storage nodes reconstruct the data blocks on the failed disk, and the MDS may also indicate A storage node acquires a data block reconstructed by each target storage node. When the data reconstruction capability of the first storage node is weak, since there are other storage nodes different from the first storage node among the n target storage nodes, other storage nodes can help the first storage node to weight the data block. Therefore, the first storage node needs less reconstructed data, and the reconstruction of the data block stored on the faulty disk is faster, so the data reconstruction efficiency of the data storage system is improved.

DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings used in the description of the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the present disclosure. Other drawings may also be obtained from those of ordinary skill in the art in light of the inventive work.

FIG. 1 is a schematic structural diagram of a data storage system according to an embodiment of the present disclosure;

2 is a schematic diagram of a disk in a storage node according to an embodiment of the present disclosure;

FIG. 3 is a flowchart of a method for data reconstruction according to an embodiment of the present disclosure;

FIG. 4 is a flowchart of another method for data reconstruction according to an embodiment of the present disclosure;

FIG. 5 is a schematic structural diagram of a data reconstruction apparatus according to an embodiment of the present disclosure;

FIG. 6 is a schematic structural diagram of another data reconstruction apparatus according to an embodiment of the present disclosure;

FIG. 7 is a schematic structural diagram of still another data reconstruction apparatus according to an embodiment of the present disclosure;

FIG. 8 is a schematic structural diagram of still another data reconstruction apparatus according to an embodiment of the present disclosure.

Detailed ways

The embodiments of the present disclosure will be further described in detail below with reference to the accompanying drawings.

FIG. 1 is a schematic structural diagram of a data storage system according to an embodiment of the present disclosure. As shown in FIG. 1 , the data storage system includes: an MDS 01, a plurality of storage nodes 02, a SAS switch 03, and an Ethernet switch 04. The MDS 01 and the plurality of storage nodes 02 are connected by an Ethernet switch 04, and the plurality of storage nodes 02 are connected by a SAS switch 03. For example, the MDS 01 may be a server or a server cluster, and the storage node 02 may be a device having a storage function, such as a server or a computer.

Each storage node consists of multiple disks, each of which is used to store data. Optionally, each storage node may further include a processor, where the storage server (English: Object Storage Device; OSD), an audit server (English: AUDITOR), and a slicing server (English: Stripe Server) Abbreviation: SS), that is, each storage node can run OSD, audit server and SS. FIG. 2 is a schematic diagram of a disk in a storage node according to an embodiment of the present disclosure. FIG. 2 shows a total of five storage nodes 02. Each storage node 02 can include multiple disks. There can be one OSD, one SS, and one audit server running on each storage storage node (neither shown in Figure 5). The OSD running on each storage node is capable of reading data stored on any disk through the SAS switch.

The user terminal can store data on the disk on the storage node in FIG. 1, and read the data in the disk on the storage node in FIG.

On the one hand, when the user terminal needs to write the target data A in the disk, the user terminal can send a write request to the MDS, at which time the MDS allocates the SS for the user terminal and assigns the EC type to the target data A. Then, the user terminal can transmit the target data A that needs to be stored to the SS allocated by the MDS. After receiving the target data A, the SS needs to apply for the stripe resource to the MDS. The MDS may allocate a strip resource to the SS according to an erasure code (English: Erasure Coding; EC) type assigned to the target data A, and the strip resource may include a disk on multiple storage nodes.

For example, the information of the stripe resource allocated by the MDS to the SS is: {<stripe_id, OSD_1, wwn_1>, <stripe_id, OSD_1, wwn_2>, <stripe_id, OSD_1, wwn_3>, <stripe_id, OSD_1, wwn_4>, <stripe_id, OSD_1, wwn_5>}. Wherein, each two adjacent "<" and ">" represents information of a disk on a storage node, and between "{" and "}" represents a stripe resource allocated for the data, and the stripe_id is a stripe ID, wwn is the disk ID. That is, the MDS allocates five disks on the five storage nodes for the target data A, namely the disks wwn_1, wwn_2, wwn_3, wwn_4, and wwn_5, and the MDS allocates the write rights of the five disks to the OSD1. These five disks are used to store the striped target data A, and the stripe identifiers in the information of the stripe resources are the same.

Then, the SS can slice the target data A according to the preset data block size (that is, the target data A is striped), and obtain k original target block target data A1 (that is, k data blocks), and according to The EC type (k+m=n) of the target data A generates m redundant target block target data A2 (that is, m data blocks) of the target data A. The stripe resources allocated by the MDS include a total of n disks on n storage nodes, and each disk is used to store one data block of the target data A.

The SS may also generate a secret key (English: key) for each data block of the target data A, the secret key of each data block may be used as an identifier of the data block, and the key of each data block may also identify the data block. The data block is the original object block target data A1 or the redundant object block target data A2. Then, the SS can obtain <stripe_id, OSD, wwn, key, value> according to the information of adding the data block and the secret key of the data block to the disk. Where value represents a data block. At this time, the information of the stripe resource becomes {<stripe_id, OSD_1, wwn_1, key_1, value_1>, <stripe_id, OSD_1, wwn_2, key_2, value_2>, <stripe_id, OSD_1, wwn_3, key_3, value_3>, <stripe_id, OSD_1, wwn_4, key_4, value_4>, <stripe_id, OSD_1, wwn_5, key_5, value_5>}.

Further, the SS may also send <wwn, key, value> in the information of each disk to the corresponding OSD, where the OSD is the OSD indicated by the OSD identifier in the information of the disk. After receiving the <wwn,key,value>, the OSD can write <key,value> to the disk indicated by wwn. After the OSD is written, the SS write success message is returned, and the SS can determine that the data block is successfully written according to the write success message. When the SS determines that each data block in the target data A (including the original object block target data A1 and the redundant object block target data A2) is successfully written, the stripe information of the target data A in the information of the strip resource may be Return to the MDS store. The stripe information of the target data A may be: {<stripe_id, wwn_1, key_1>, <stripe_id, wwn_2, key_2>, <stripe_id, wwn_3, key_3>, <stripe_id, wwn_4, key_4>, <stripe_id, wwn_5, Key_5>}. Wherein, between "{" and "}" is represented as stripe information of the target data A, and each two adjacent "<" and ">" represents storage information of one data block, and stripe of the target data A The information is also referred to as stripe information of each data block of the target data A.

On the other hand, when the user terminal needs to read the target data A stored in the disk, the user terminal needs to send a read request to the MDS. The MDS can read the stripe information of the previously recorded target data A according to the read request. For example, the stripe information of the target data A is {<stripe_id, wwn_1, key_1>, <stripe_id, wwn_2, key_2>, <stripe_id, wwn_3, key_3>, <stripe_id, wwn_4, key_4>, <stripe_id, wwn_5, key_5> }. Then, the MDS may determine the OSD that previously stored the target data A according to the disk indicated by wwn in the strip information, and send the stripe information of the target data A and the identifier of the user terminal to the SS of the OSD local (the OSD) Running on the same storage node as the SS).

After receiving the stripe information of the target data A, the SS may send <wwn, key> in the storage information of each original target block target data A1 in the stripe information to the local OSD. Then, the OSD can read the data block (also called value) on the disk indicated by wwn according to the received key in <wwn, key>, and return <key, value> to the local SS. After receiving all <key, value> returned by the OSD, the SS can get the received <key, value> combination to get {<key_1, value_1>, <key_2, value_2>, <key_3, value_3>} (assuming value_1, Both value_2 and value_3 are the original object block target data A1, and value_4 and value_5 are redundant object block target data A2). Finally, the SS can package the value_1, value_2, and value_3 in {<key_1, value_1>, <key_2, value_2>, <key_3, value_3>} to obtain the target data A, and send the target data A to the user terminal.

It should be noted that the disk in the storage node is more likely to be faulty, and the user cannot read the data block stored on the faulty disk. Therefore, the embodiment of the present disclosure provides a data reconstruction method for Reconstruct the data blocks stored on the failed disk.

FIG. 3 is a flowchart of a method for data reconstruction according to an embodiment of the present disclosure. The data reconstruction method may be used for an MDS in a data storage system (such as MDS 01 shown in FIG. 1 ), as shown in FIG. 3 . The data reconstruction method includes:

Step 301: When detecting a faulty disk in which the m data blocks are stored in the first storage node, send a reconstruction instruction to each of the plurality of storage nodes, where the first storage node is multiple Any one of the storage nodes, the n reconstruction instructions sent to the n target reconstruction nodes are used to indicate that the m data blocks are reconstructed and stored, and the n storage nodes exist and the first storage node Different storage nodes, m≥n≥1.

Step 302: Send an acquisition instruction to the first storage node, where the acquisition instruction is used to instruct the first storage node to acquire and store the m data blocks reconstructed by the n target storage nodes.

In summary, the embodiment of the present disclosure provides a data reconstruction method. When detecting a faulty disk in a first storage node, the MDS sends a reconstruction instruction to the n target storage nodes, so that n target storages are performed. The node reconstructs the data block on the failed disk, and the MDS may also instruct the first storage node to acquire the data block reconstructed by each target storage node. When the data reconstruction capability of the first storage node is weak, since there are other storage nodes different from the first storage node among the n target storage nodes, other storage nodes can help the first storage node to weight the data block. Therefore, the first storage node needs less reconstructed data, and the reconstruction of the data block stored on the faulty disk is faster, so the data reconstruction efficiency of the data storage system is improved.

FIG. 4 is a flowchart of another method for data reconstruction according to an embodiment of the present disclosure. As shown in FIG. 6, the data reconstruction method includes:

Step 401: The first storage node sends a fault message to the MDS, where the fault message is used to indicate that a faulty disk is present in the first storage node.

It should be noted that each storage node in FIG. 1 may include multiple disks, and the multiple disks include a storage disk and a cache disk. The storage disk and the cache disk may be solid state drives (English: Solid State Drives; SSD), serial hard disks (also called SATA hard disks), or SAS disks. Optionally, the cache disk is a solid state drive (English: Solid State Drives; SSD), and the storage disk is a serial hard disk (also called a SATA hard disk) or a SAS disk. The first storage node is any one of the plurality of storage nodes, and the failed disk may be a storage disk of the first storage node.

Referring to FIG. 2, a plurality of disks (six disks shown in FIG. 2) of each storage node 02 may include: five storage disks and one cache disk. The OSD running on each storage node 02 has the right to write data to the cache disk on the storage node 02, and has the right to read data for each storage disk and each cache disk on the storage node 02. It should be noted that the number of the storage disks in the storage node 02 may be any integer greater than or equal to 1, and the number of the cache disks may be any integer greater than or equal to 1, which is not limited in the embodiment of the present disclosure. Optionally, the OSD can be used to monitor whether the storage disk in the storage node is faulty. When a storage disk fails, the OSD determines that the storage disk is a faulty disk, and sends a fault message to the MDS, and the fault message can be used to indicate the fault. Faulty disk. For example, the fault message can include an identification of the failed disk.

Step 402: The MDS determines a faulty disk in the first storage node according to the fault message.

After receiving the fault message, the MDS can parse the fault message, obtain the identifier of the faulty disk in the fault message, and determine the faulty disk in the first storage node.

Step 403: The MDS acquires stripe information of m data blocks stored on the faulty disk.

It should be noted that the MDS stores strip information of each data block stored on each disk in the data storage system. After the MDS determines the faulty disk in the first storage node, the MDS may determine m data blocks stored on the faulty disk, and obtain stripe information of each of the m data blocks. Where m≥1.

Step 404: The MDS determines n target storage nodes in the data storage system.

For example, m≥n≥1, that is, the MDS may select one storage node as the target storage node in the data storage system, or select a plurality of storage nodes as the target storage node in the data storage system. And when n=1, a target storage node selected by the MDS is not the first storage node, and when n≥2, the plurality of target storage nodes selected by the MDS may include the first storage node, or may not include the first Storage node.

On the one hand, when n = 1, the MDS can determine a target storage node in the data storage system in a variety of implementable ways. The following three implementable modes will be exemplified in the embodiments of the present disclosure.

In a first implementation, the MDS may first determine the load of each storage node in the data storage system other than the first storage node. It should be noted that the load of the storage node may be positively correlated with at least one performance parameter of the storage node, where the performance parameters of the storage node include: usage of the processor in the storage node, memory of the storage node (including all in the storage node) Disk usage and storage efficiency of storage nodes.

The MDS can determine a target storage node based on the load of each storage node other than the first storage node. For example, the MDS may compare the load of the storage nodes other than the first storage node in the data storage system, and determine one storage node with the smallest load as the target storage node. That is, the MDS may select a storage node with a minimum load (higher data processing capability) other than the first storage node as a target for performing a data reconstruction task when selecting a target storage node that needs to perform a reconstruction task. The storage node is configured to ensure that the target storage node can perform the task of reconstructing the data block faster, and improve the efficiency of data reconstruction.

In a second implementation manner, a storage node in the data storage system has a preset storage node different from the first storage node, and the preset storage node may be a storage node with higher data processing capability. The MDS can directly determine the preset storage node as the target storage node.

In a third implementation manner, the MDS may first determine a routing distance of each storage node except the first storage node and the MDS in the data storage system, and according to each storage node except the first storage node. The routing distance from the MDS determines a target storage node. For example, the MDS may compare the storage node of the data storage system except the first storage node with the routing distance of the MDS, and determine a storage node with the smallest routing distance from the MDS as the target storage node. That is, when selecting the target storage node that needs to perform the reconstruction task, the MDS may select the storage node closest to the routing distance of the MDS as the storage data reconstruction task among the storage nodes other than the first storage node. The target storage node ensures that the MDS can quickly allocate the reconstructed data block to the target storage node, thereby improving the efficiency of data reconstruction.

On the other hand, when n ≥ 2, the MDS can determine n target storage nodes in the data storage system in a variety of implementable ways. The following three implementable modes will be exemplified in the embodiments of the present disclosure.

In a first implementation, the MDS may first determine the load of each storage node in the data storage system. The MDS can determine n target storage nodes based on the load of each storage node. For example, a preset number threshold n may be pre-stored on the MDS. The MDS can compare the loads of the storage nodes in the data storage system and determine the n storage nodes with smaller loads as the n target storage nodes. That is, when the MDS selects the target storage node that needs to perform the reconstruction task, the storage node with smaller load (higher data processing capability) can be selected as the target storage node for performing the data reconstruction task to ensure the target storage. The node can perform the task of reconstructing the data block faster, and improve the efficiency of data reconstruction.

In the second implementation manner, there are n preset storage nodes in the storage node in the data storage system, and the n preset storage nodes may be storage nodes with higher data processing capability. The MDS can directly determine the preset n storage nodes as n target storage nodes.

In a third implementation manner, the MDS may first determine a routing distance between each storage node and the MDS in the data storage system, and determine n target storage nodes according to a routing distance between each storage node and the MDS. For example, a preset number threshold n may be pre-stored on the MDS. The MDS can compare the storage nodes of the data storage system with the routing distance of the MDS, and determine n storage nodes with a small routing distance from the MDS as n target storage nodes. That is, when selecting the target storage node that needs to perform the reconstruction task, the MDS may select the storage node that is closer to the routing distance of the MDS as the target storage node for performing the data reconstruction task, so as to ensure that the MDS can be quickly followed. The task of allocating reconstructed data blocks to each target storage node improves the efficiency of data reconstruction.

Step 405: The MDS separately sends n reconstruction instructions to the n target storage nodes.

After determining the stripe information of each data block in the m data blocks and the n target storage nodes, the MDS may determine the correspondence of each target storage node according to the load of the n target storage nodes and the stripe information of the m data blocks. At least one data block. The sum of the data amounts of all the data blocks corresponding to the target storage node is negatively correlated with the load of the target storage node.

If the load of the target storage node is large, the target storage node can reconstruct data with a smaller amount of data. At this time, the sum of the data amounts of all the data blocks corresponding to the target storage node is small; if the load of the target storage node is small The target storage node can reconstruct data of a larger amount of data. At this time, the sum of the data amounts of all the data blocks corresponding to the target storage node is large. That is, the data reconstruction capability of the target storage node is related to the load of the target storage node. The MDS needs to allocate a data block to be reconstructed for each target storage node according to the load and reconstruction capability of each target storage node.

After determining the data block corresponding to each target storage node, the MDS may generate a reconstruction instruction corresponding to each target storage node according to the stripe information of the data block. The reconfiguration instruction corresponding to each target storage node is used to indicate that the data block corresponding to each target storage node is reconstructed and stored. Afterwards, the MDS can send the reconfiguration instruction corresponding to each target storage node to the audit server running on each target storage node.

For example, the reconfiguration instruction corresponding to each target storage node includes: stripe information of each data block corresponding to each target storage node, and a storage node for indicating whether each target storage node is a faulty disk. Instructions.

Step 406: The n target storage nodes reconstruct and store the data block according to the received reconstruction instruction.

After receiving the reconstruction instruction, the audit server running on each target storage node may parse the reconstruction instruction to obtain stripe information of at least one data block that needs to be reconstructed.

The audit server running on each target storage node can also read the valid data required for reconstructing each data block through the local OSD according to the stripe information of each data block that needs to be reconstructed, and the valid data storage. In at least one disk in the data storage system. It should be noted that, assuming that the data block X is a data block that needs to be reconstructed, and the stripe information of the data block X includes: the storage information of the data block X, and the storage information of the data block Y, the process of reconstructing the data block X The valid data required in is the data block Y. Afterwards, the audit server running on each target storage node can reconstruct the corresponding data block according to the valid data read and the received reconstruction instructions.

Further, after receiving the reconfiguration instruction, the audit server running on each target storage node may parse the reconfiguration instruction to obtain a storage node for indicating whether each target storage node is a faulty disk. Instructions.

In a first aspect, when the indication information is used to indicate that the target storage node is not the storage node where the faulty disk is located, the audit server running on the target storage node may determine that the reconstruction instruction is used to indicate that the target storage node is to be reconstructed. The data block is stored on the cache disk. After the audit server running on the target storage node reconstructs the data block, the audit server running on the target storage node may send the reconstructed data block to the local OSD and instruct the local OSD to write the data block to the target. The cache disk in the storage node.

In a second aspect, when the indication information is used to indicate that the target storage node is a storage node (ie, the first storage node) where the faulty disk is located, the audit server running on the first storage node may determine that the reconstruction instruction is used to indicate the The first storage node stores the reconstructed data block on the cache disk. After the audit server running on the first storage node reconstructs the data block, the audit server running on the first storage node may send the reconstructed data block to the local OSD and instruct the local OSD to write the data block. The first storage node is in the cache disk.

In a third aspect, when the indication information is used to indicate that the target storage node is a storage node where the faulty disk is located (that is, the first storage node), the audit server running on the first storage node may also determine that the reconstruction instruction is used to indicate The target storage node stores the reconstructed data block on a storage disk. The audit server running on the first storage node may send a storage request message to the MDS, the storage request message including the total data amount of the m data blocks. The MDS may send, according to the storage request message, a storage instruction to the audit server running on the first storage node, where the storage instruction is used to instruct the first storage node to store the reconstructed m data blocks on the target disk, where the target disk may be a storage disk in which the storage capacity of the first storage node is greater than or equal to the total data amount of the m data blocks, and the related data block is not stored on the target disk, and the related data block and any data in the m data blocks are not stored. Blocks belong to the same strip. After the audit server running on the first storage node reconstructs the data block, the audit server running by the first storage node may send the reconstructed data block to the local OSD, and instruct the local OSD to write the data block to the first In the target disk in the storage node.

Step 407: The n target storage nodes respectively send a reconstruction complete message to the MDS.

After each target storage node reconstructs all the data blocks according to the reconstruction instruction and stores all the reconstructed data blocks, the target storage node may send a reconstruction completion message to the MDS. The reconstruction completion message sent by each target storage node may include: an identifier of each data block reconstructed by the target storage node, and an identifier of a disk stored by each of the data blocks.

Step 408: The MDS sends an acquisition instruction to the first storage node, where the acquisition instruction is used to instruct the first storage node to acquire and store the m data blocks reconstructed by the n target storage nodes.

After the MDS receives the reconstruction completion message sent by the n target storage nodes, the MDS may determine that each target storage node has completed the data reconstruction task of the MDS allocation. At this time, the MDS may run on the first storage node. The audit server sends an acquisition instruction to instruct the audit server to acquire and store the data block reconstructed by each target storage node. It should be noted that the obtaining instruction may include: an identifier of each data block in the m data blocks, and an identifier of a disk stored in each of the data blocks.

Step 409: The first storage node acquires and stores the reconstructed m data blocks according to the acquisition instruction.

The audit server running on the first storage node may determine the identifier of the disk stored in each of the reconstructed m data blocks according to the obtaining instruction, and obtain the reconstructed data block on the corresponding disk.

For example, when the disk stored in the reconstructed data block is not the disk in the first storage node, the audit server running on the first storage node can read or copy the corresponding disk through the local OSD and SAS switches. Reconstructed data blocks stored on (cache disks of other storage nodes). When the disk stored in the reconstructed data block is a cache disk in the first storage node, the audit server running on the first storage node can directly read the reconstructed data block stored on the local cache disk. When the disk stored in the reconstructed data block is the target disk in the first storage node, the audit server running on the first storage node can obtain the reconstructed data block without performing the step of reading the data block. .

After the reconstructed data block is obtained, the audit server running on the first storage node may store the reconstructed data block.

For example, when the reconstructed data block is implemented by the first aspect or the second aspect in step 406, in step 409, the audit server running on the first storage node may further send a storage application message to the MDS, the storage application. The message can include the total amount of data for m data blocks. The MDS may send, according to the storage request message, a storage instruction to the audit server running on the first storage node, where the storage instruction is used to instruct the first storage node to store the reconstructed m data blocks on the target disk. Then, the audit server running on the first storage node can store the obtained reconstructed m data blocks on the target disk.

When the reconstructed data block is implemented by the third aspect in step 406, the audit server running on the first storage node may directly store the obtained reconstructed m data blocks on the target disk in step 409, and The step of performing repeated storage on the reconstructed data block stored by the first storage node is performed, and the reconstructed data block can be guaranteed to be stored on the target disk.

Step 410: The first storage node sends an acquisition completion message to the MDS.

After obtaining the reconstructed data block, the audit server running on the first storage node may send an acquisition completion message to the MDS, where the acquisition completion message may be used to indicate that the first storage node has acquired the reconstruction of each target storage node. data block.

Step 411: The first storage node sends, to the MDS, storage information of each of the m data blocks, where the storage information of each data block includes: an identifier of the target disk and an identifier of each data block.

After the audit server running on the first storage node determines that each reconstructed data block is written to the target disk, the storage information of each data block may be sent to the MDS, where the storage information includes the identifier of the data block (such as the data block). Key), and the identity of the target disk where the data block is located.

Step 412: The MDS updates the stripe information of each data block in the m data blocks.

After receiving the storage information of each data block, the MDS may search for the strip information of each data block according to the identifier of the data block in the storage information of each data block, and the strip information of each data block is The ID of the disk where the data block is located is modified to the ID of the target disk.

For example, the identifier of the data block X is key1. Before modifying the identifier of the disk in the stripe information, the stripe information of the data block X may be: {<stripe_id, wwn_1, key_1>, <stripe_id, wwn_2, key_2>, <stripe_id, wwn_3, key_3>, <stripe_id, wwn_4, key_4>, <stripe_id, wwn_5, key_5>}, in step 413, the MDS can modify wwn_1 in the stripe information to wwn_x (identification of the target disk), Thus, the stripe information of the data block X is updated to: {<stripe_id, wwn_x, key_1>, <stripe_id, wwn_2, key_2>, <stripe_id, wwn_3, key_3>, <stripe_id, wwn_4, key_4>, <stripe_id, wwn_5, Key_5>}.

Step 413: The MDS sends a delete instruction to each of the n target storage nodes, where the delete instruction is used to instruct the cache storage node to delete the data block stored on the cache disk of the cache storage node.

It should be noted that the n target storage nodes include cache storage nodes, and each cache storage node stores the reconstructed data blocks on the cache disk of the cache storage node after reconstructing the data block. For example, when the first aspect or the second aspect is implemented in step 406, each target storage node is a cache storage node. When the third aspect is implemented in step 406, the first storage node is excluded from the n target storage nodes. Each target storage node is a cache storage node.

After receiving the acquisition completion message sent by the audit server running on the first storage node, the MDS may send a delete instruction to the OSD running on each cache storage node to indicate that the OSD running on each cache storage node deletes the cache storage. The data block (that is, the reconstructed data block) stored on the node's cache disk.

Step 414: Each cache storage node deletes a data block stored on a cache disk of the cache storage node according to the delete instruction.

After receiving the delete instruction, the OSD running on each cache storage node can directly delete the data block stored on the cache disk of the cache storage node.

For example, as shown in FIG. 2, assuming that the first storage node is the storage node 1, and the failed disk is the storage disk 1-1 on the first storage node, the MDS may determine the storage node 1, the storage node 2, and the storage node 3. The storage node 4 and the storage node 5 are both target storage nodes.

The MDS can also send reconstruction instructions to the storage node 1, the storage node 2, the storage node 3, the storage node 4, and the storage node 5, respectively. The storage node 1 can reconstruct the data block 1 according to the received reconstruction instruction, the storage node 2 can reconstruct the data block 2 according to the received reconstruction instruction, and the storage node 3 can reconstruct the data block 3 according to the received reconstruction instruction. The storage node 4 can reconstruct the data block 4 according to the received reconstruction instruction, and the storage node 5 can reconstruct the data block 5 according to the received reconstruction instruction. It should be noted that the storage disk 1-1 (faulty disk) stores the data block 1, the data block 2, the data block 3, the data block 4, and the data block 5.

The storage node 1 may also send a storage request message to the MDS, and the MDS may send a storage instruction to the storage node 1 for instructing the storage node 1 to store the data block on the storage disk 6-1 (target disk). The storage node 1 can store the reconstructed data block 1 on the storage disk 6-1, the storage node 2 can store the reconstructed data block 2 on the cache disk 2, and the storage node 3 can store the reconstructed data block 3 On the cache disk 3, the storage node 4 can store the reconstructed data block 4 on the cache disk 4, and the storage node 5 can store the reconstructed data block 5 on the cache disk 5.

After storing the reconstructed data block, each target storage node may send a reconstruction complete message to the MDS, so that the MDS sends an acquisition instruction to the storage node 1 after receiving the reconstructed message sent by all the target storage nodes. The storage node 1 can obtain the reconstructed data block 2, the data block 3, the data block 4, and the data block stored on the cache disk 2, the cache disk 3, the cache disk 4, and the cache disk 5 through the SAS switch according to the received acquisition instruction. 5, and the data block 2, the data block 3, the data block 4, and the data block 5 are also stored on the storage disk 6-1.

After the storage node 1 obtains the data block reconstructed by each target storage node, the storage node 1 may also send an acquisition complete message to the MDS, and the MDS may store the storage node 2, the storage node 3, and the storage according to the received acquisition completed message. The node 4 and the storage node 5 respectively send a delete instruction to instruct the storage node 2, the storage node 3, the storage node 4, and the storage node 5 to delete the data blocks stored on the local cache disk, respectively. After the storage node 1 stores the data block on the storage disk 6-1, the storage node 1 can also transmit the storage information of the data block like the MDS. The MDS can update the stripe information of the data block according to the storage information of the data block.

FIG. 5 is a schematic structural diagram of a data reconstruction apparatus according to an embodiment of the present disclosure. The data reconstruction apparatus may be used in an MDS (such as the MDS shown in FIG. 1) in a data storage system, as shown in FIG. The data reconstruction device 50 can include:

a first sending module 501, configured to send a reconfiguration instruction to each of the plurality of storage nodes when detecting a faulty disk in which the m data blocks are stored in the first storage node, where the first The storage node is any one of the plurality of storage nodes, and the n reconstruction instructions sent to the n target reconstruction nodes are used to indicate that the m data blocks are reconstructed and stored, and the n target storage nodes exist. a storage node different from the first storage node, m≥n≥1;

The second sending module 502 is configured to send an acquisition instruction to the first storage node, where the obtaining instruction is used to instruct the first storage node to acquire and store the m data blocks reconstructed by the n target storage nodes.

In summary, the embodiment of the present disclosure provides a data reconstruction apparatus, where the first sending module sends a reconfiguration instruction to the n target storage nodes when detecting a faulty disk in the first storage node, so that n The target storage node reconstructs the data block on the failed disk, and the second sending module may instruct the first storage node to acquire the data block reconstructed by each target storage node. When the data reconstruction capability of the first storage node is weak, since there are other storage nodes different from the first storage node among the n target storage nodes, other storage nodes can help the first storage node to weight the data block. Therefore, the first storage node needs less reconstructed data, and the reconstruction of the data block stored on the faulty disk is faster, so the data reconstruction efficiency of the data storage system is improved.

Optionally, m≥n≥2, the data reconstruction device 50 may further include:

a first determining module (not shown in FIG. 5), configured to determine n storage nodes with smaller loads among the plurality of storage nodes as n target storage nodes;

Or a second determining module (not shown in FIG. 5), configured to determine n storage nodes preset among the plurality of storage nodes as n target storage nodes;

Alternatively, the third determining module (not shown in FIG. 5) is configured to determine n storage nodes that are closest to the routing distance of the MDS among the plurality of storage nodes as n target storage nodes.

Optionally, the first sending module 501 is configured to: determine a load of each target storage node; determine, according to a load of the n target storage nodes, at least one data block corresponding to each target storage node, where the target storage node corresponds to The sum of the data amounts of all the data blocks is negatively correlated with the load of the target storage node; generating a reconstruction instruction corresponding to each target storage node, wherein the reconstruction instruction corresponding to each target storage node is used to indicate: for each target The data block corresponding to the storage node is reconstructed and stored; and each corresponding storage node is sent its corresponding reconstruction instruction.

Optionally, FIG. 6 is a schematic structural diagram of another data reconstruction apparatus according to an embodiment of the present disclosure. As shown in FIG. 6, the data reconstruction apparatus 50 may further include:

The first receiving module 503 is configured to receive a storage application message sent by the first storage node, where the storage application message includes a total data volume of the m data blocks;

The third sending module 504 is configured to send, to the first storage node, a storage instruction according to the storage request message, where the storage instruction is used to store the reconstructed m data blocks on the target disk, where the target disk is an available storage capacity in the first storage node. A disk that is greater than or equal to the total amount of data.

Optionally, the relevant data block is not stored on the target disk, and the related data block belongs to the same strip as any one of the m data blocks.

Optionally, FIG. 7 is a schematic structural diagram of another data reconstruction apparatus according to an embodiment of the present disclosure. As shown in FIG. 7, the data reconstruction apparatus 50 may further include:

The second receiving module 505 is configured to receive storage information of each data block sent by the first storage node, where the storage information of each data block includes: an identifier of the target disk and an identifier of each data block;

a fourth determining module 506, configured to determine strip information of each data block according to an identifier of each data block in the storage information of each data block;

The modifying module 507 is configured to modify the identifier of the disk where each data block in each strip of the data block is located to be the identifier of the target disk.

Optionally, each of the plurality of storage nodes includes: a storage disk and a cache disk, each storage node has read permission of the storage disk, and read and write permissions of the cache disk, and the n target storage nodes include a cache storage node. The reconfiguration instruction sent to the cache storage node is used to indicate that the reconstructed data block is stored in the cache disk of the cache storage node. FIG. 8 is a schematic structural diagram of another data reconstruction apparatus according to an embodiment of the present disclosure, such as As shown in FIG. 8, on the basis of FIG. 5, the data reconstruction apparatus 50 may further include:

The third receiving module 508 is configured to receive the obtained completion message sent by the first storage node, where the obtained complete message is used to indicate that the first storage node has acquired and stored the reconstructed m data blocks.

The fourth sending module 509 is configured to send a delete instruction to the cache storage node, where the delete instruction is used to instruct the cache storage node to delete the data block stored on the cache disk.

Embodiments of the present disclosure provide a computer device having a computer program running thereon, the processor in the computer device executing a computer program to implement the data reconstruction method described above. The MDS in the data storage system shown in Figure 1 can include the computer device.

Embodiments of the present disclosure provide a storage medium on which a computer program is stored, and a processor executes a computer program to implement the data reconstruction method described above.

Embodiments of the present disclosure provide a computer program product that, when executed on a computer, causes the computer to perform the data reconstruction method described above.

It should be noted that the embodiment of the method provided by the embodiment of the present disclosure can refer to the corresponding device embodiment, and the embodiment of the present disclosure does not limit this. The sequence of the steps of the method embodiments provided by the embodiments of the present disclosure can be appropriately adjusted, and the steps can also be correspondingly increased or decreased according to the situation. Any person skilled in the art can easily think of changes within the technical scope disclosed by the disclosure. The method should be covered by the scope of the present disclosure, and therefore will not be described again.

A person skilled in the art may understand that all or part of the steps of implementing the above embodiments may be completed by hardware, or may be instructed by a program to execute related hardware, and the program may be stored in a computer readable storage medium. The storage medium mentioned may be a read only memory, a magnetic disk or an optical disk or the like.

The above description is only an alternative embodiment of the present disclosure, and is not intended to limit the disclosure, and any modifications, equivalents, improvements, etc., made within the spirit and principles of the present disclosure should be included in the protection of the present disclosure. Within the scope.

Claims

A data reconstruction method for a metadata management server MDS in a data storage system, the data storage system further comprising: a serial connection small computer system interface SAS switch and a plurality of storage nodes, wherein the plurality of storage nodes pass The SAS switches are connected to each other, and the method includes:

When detecting the faulty disk in which the m data blocks are stored in the first storage node, respectively sending a reconstruction instruction to the n target storage nodes of the plurality of storage nodes, where the first storage node is ???said one of the plurality of storage nodes, and the n reconstruction instructions sent to the n target reconstruction nodes are used to indicate that the m data blocks are reconstructed and stored, the n targets There is a storage node different from the first storage node in the storage node, m≥n≥1;

Sending an acquisition instruction to the first storage node, where the obtaining instruction is used to instruct the first storage node to acquire and store the m data blocks reconstructed by the n target storage nodes.
The method of claim 1, wherein m≥n≥2, before separately transmitting the reconstruction instruction to the n target storage nodes of the plurality of storage nodes, the method further comprises:

Determining, as the n target storage nodes, n storage nodes having a smaller load among the plurality of storage nodes;

Or determining, by the n storage nodes preset in the plurality of storage nodes, the n target storage nodes;

Alternatively, the n storage nodes that are closest to the routing distance of the MDS among the plurality of storage nodes are determined as the n target storage nodes.
The method according to claim 1 or 2, wherein the reconfiguring instructions are respectively sent to the n target storage nodes of the plurality of storage nodes, including:

Determining a load of each of the target storage nodes;

Determining, according to the load of the n target storage nodes, at least one data block corresponding to each target storage node, wherein a sum of data amounts of all data blocks corresponding to the target storage node is negatively correlated with a load of the target storage node;

Reconstructing the reconstruction instruction corresponding to each target storage node, where the reconstruction instruction corresponding to each target storage node is used to indicate that the data block corresponding to each target storage node is reconstructed and stored;

Sending its corresponding reconstruction instruction to each of the target storage nodes.
The method of claim 1, after the resizing instructions are respectively sent to the n target storage nodes of the plurality of storage nodes, the method further comprising:

Receiving a storage application message sent by the first storage node, where the storage application message includes a total data amount of the m data blocks;

And sending, by the storage request message, a storage instruction to the first storage node, where the storage instruction is used to indicate that the reconstructed m data blocks are stored on a target disk, where the target disk is the first storage node A disk in which the available storage capacity is greater than or equal to the total amount of data.
The method of claim 4,

An associated data block is not stored on the target disk, and the related data block belongs to the same strip as any one of the m data blocks.
The method of claim 4 or 5, after the sending the storage instruction to the first storage node, the method further comprises:

Receiving, by the first storage node, the storage information of each of the data blocks, where the storage information of each data block includes: an identifier of the target disk and an identifier of each of the data blocks;

Determining stripe information of each of the data blocks according to the identifier of each of the data blocks in the storage information of each data block;

Modifying, in the stripe information of each data block, an identifier of a disk where each of the data blocks is located, to an identifier of the target disk.
The method of claim 1, each of the plurality of storage nodes comprising: a storage disk and a cache disk, each storage node having read access to the storage disk, and the cache disk Read and write rights, the n target storage nodes include a cache storage node, and the reconstruction instruction sent to the cache storage node is used to indicate that the reconstructed data block is stored in a cache disk of the cache storage node, After the first storage node sends the acquisition instruction, the method further includes:

And receiving the obtained completion message sent by the first storage node, where the obtained information is used to indicate that the first storage node has acquired and stored the reconstructed m data blocks;

And sending a delete instruction to the cache storage node, where the delete instruction is used to instruct the cache storage node to delete a data block stored on a cache disk thereof.
A data reconstruction apparatus is used for a metadata management server MDS in a data storage system, the data storage system further comprising: a serial connection small computer system interface SAS switch and a plurality of storage nodes, wherein the plurality of storage nodes pass The SAS switches are connected to each other, and the method includes:

a first sending module, configured to send a reconfiguration instruction to each of the plurality of storage nodes when detecting a faulty disk in which the m data blocks are stored in the first storage node, where The first storage node is any one of the plurality of storage nodes, and the n reconstruction instructions sent to the n target reconstruction nodes are used to indicate that the m data blocks are reconstructed and Storage, a storage node different from the first storage node exists in the n target storage nodes, m≥n≥1;

a second sending module, configured to send an acquisition instruction to the first storage node, where the obtaining instruction is used to instruct the first storage node to acquire and store the m data blocks reconstructed by the n target storage nodes .
The data reconstruction device according to claim 8, wherein m≥n≥2, the data reconstruction device further comprises:

a first determining module, configured to determine n storage nodes that are less loaded among the plurality of storage nodes as the n target storage nodes;

Or the second determining module is configured to determine, as the n target storage nodes, the n storage nodes preset in the multiple storage nodes;

Alternatively, the third determining module is configured to determine, as the n target storage nodes, n storage nodes that are closest to the routing distance of the MDS among the plurality of storage nodes.
The data reconstruction device according to claim 8 or 9, wherein the first sending module is configured to:

Determining a load of each of the target storage nodes;

Determining, according to the load of the n target storage nodes, at least one data block corresponding to each target storage node, wherein a sum of data amounts of all data blocks corresponding to the target storage node is negatively correlated with a load of the target storage node;

Reconstructing the reconstruction instruction corresponding to each target storage node, where the reconstruction instruction corresponding to each target storage node is used to indicate that the data block corresponding to each target storage node is reconstructed and stored;

Sending its corresponding reconstruction instruction to each of the target storage nodes.
The data reconstruction device according to claim 8, wherein the data reconstruction device further comprises:

a first receiving module, configured to receive a storage application message sent by the first storage node, where the storage application message includes a total data volume of the m data blocks;

a third sending module, configured to send a storage instruction to the first storage node according to the storage request message, where the storage instruction is used to indicate that the reconstructed m data blocks are stored on a target disk, the target disk A disk that has a storage capacity greater than or equal to the total amount of data in the first storage node.
The data reconstruction device according to claim 11,

An associated data block is not stored on the target disk, and the related data block belongs to the same strip as any one of the m data blocks.
The data reconstruction device according to claim 11 or 12, wherein the data reconstruction device further comprises:

a second receiving module, configured to receive storage information of each of the data blocks sent by the first storage node, where the storage information of each data block includes: an identifier of the target disk, and each of the data blocks Identification

a fourth determining module, configured to determine stripe information of each of the data blocks according to the identifier of each of the data blocks in the storage information of each data block;

And a modification module, configured to modify an identifier of the disk where each of the data blocks in the strip information of each data block is located to an identifier of the target disk.
The data reconstruction device of claim 8, each of the plurality of storage nodes comprising: a storage disk and a cache disk, each storage node having read permission of the storage disk, and the The read and write permissions of the cache disk, the n target storage nodes include a cache storage node, and the reconstruction instruction sent to the cache storage node is used to indicate: storing the reconstructed data block in a cache disk of the cache storage node The data reconstruction device further includes:

a third receiving module, configured to receive the acquired information message sent by the first storage node, where the obtained message is used to indicate that the first storage node has acquired and stored the reconstructed m data blocks;

And a fourth sending module, configured to send a delete instruction to the cache storage node, where the delete instruction is used to instruct the cache storage node to delete a data block stored on a cache disk.
A data storage system, comprising: a metadata management server MDS, a plurality of storage nodes, and a serial connection small computer system interface SAS switch, wherein the plurality of storage nodes are connected to each other through the SAS switch, The MDS comprises the data reconstruction device of any of claims 8 to 14.
A computer device comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface, the memory complete communication with each other through the bus; the memory for storing the computer program; and the processor for executing the memory The stored program implements the data reconstruction method of any one of claims 1-7.
A data storage system, comprising: a metadata management server MDS, a plurality of storage nodes, and a serial connection small computer system interface SAS switch, wherein the plurality of storage nodes are connected to each other through the SAS switch, The MDS comprises the computer device of claim 16.
A computer readable storage medium having stored therein a computer program, the computer program being executed by a processor to implement the data reconstruction method of any of claims 1-7.