WO2024098885A1

WO2024098885A1 - Disk array write processing method and apparatus, device, and medium

Info

Publication number: WO2024098885A1
Application number: PCT/CN2023/114470
Authority: WO
Inventors: 梁欣玲; 邸忠辉
Original assignee: 苏州元脑智能科技有限公司
Priority date: 2022-11-11
Filing date: 2023-08-23
Publication date: 2024-05-16
Also published as: CN115469818B; CN115469818A

Abstract

The present application discloses a disk array write processing method and apparatus, a device, and a medium, which are applied to the technical field of data storage. The method comprises: on the basis of an IO write request, determining a target stride to which data is to be written in DRAID5 and write location information of a target data block to which data is to be written in the target stride; respectively recording the write location information of the target data block in each target stride to a corresponding information record; on the basis of the information record, determining total location information, of a check block in each target stride, on which data updating is to be performed; when any target stride meets a check data reading condition, performing primary data reading from a corresponding check block on the basis of current total location information so as to obtain old check data corresponding to the current total location information; performing calculation on the basis of the old check data to obtain new check data; and writing the new check data to the corresponding check block on the basis of the total location information. The time spent processing check data can be reduced, so that the storage performance is improved.

Description

Disk array write processing method, device, equipment and medium

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims priority to a Chinese patent application filed with the Chinese Patent Office on November 11, 2022, with application number 202211412299.2, and entitled “A disk array write processing method, device, equipment and medium”, all contents of which are incorporated by reference in this application.

Technical Field

The present application relates to the technical field of data storage, and in particular to a disk array write processing method, device, equipment and medium.

Background technique

In the information age with rapid technological development, performance has gradually become one of the main criteria for storage evaluation. DRAID5 (Redundant Array of Independent Disks) is a common disk array, which is composed of strides, which are composed of strips (data blocks and check blocks). To ensure the consistency of stripes, a check block in a stripe is obtained by XORing all data blocks in the stripe.

Currently, when the server initiates a write IO (Input/Output) process and writes data to different data blocks, it is necessary to read the data in the stripe check block where each data block is located based on the offset and length, calculate the new check data, and write the new check data. When the write process is initiated, there may be tens of thousands of IOs per second, which will take a lot of time and affect storage performance.

Summary of the invention

In view of this, the purpose of this application is to provide a disk array write processing method, device, equipment and medium, which can reduce the time consumption of processing verification data, thereby improving storage performance. The specific scheme is as follows:

The present application discloses a disk array write processing method, comprising:

Determine the target stripe to be written to data in DRAID5 and the write position information of the target data block to be written to data in each target stripe based on the IO write request initiated by the server; the write position information includes the write offset and the write length;

Recording the write position information of the target data block in each target stripe to the information record corresponding to the target stripe;

Determine the full amount of location information of the parity block in each target stripe that is currently to be updated based on the information record;

When any target stripe meets the check data reading condition, data is read from the corresponding check block based on the current full position information to obtain the old check data corresponding to the current full position information;

Perform calculation based on old verification data to obtain new verification data;

Write the new verification data into the corresponding verification block based on the full location information.

Optionally, the full amount of location information of the parity block in each target stripe that is currently to be updated with data is determined based on the information record, including:

The write position information is read from the information record in sequence, and each time a write position information is read, the write position information is converted into a page mask form to obtain a page mask corresponding to the write position information, and the page mask is ORed with the target page mask to obtain a position information mask corresponding to the write position information, and the position information mask is recorded as the full amount of position information of the parity block in the target stripe that is currently to be updated with data;

Among them, the page mask format is a binary code format of a preset number of bits, each bit corresponds to a page, and the target page mask corresponding to the first write position information read is the initial page mask, and the target page mask corresponding to the non-first write position information is the position information mask corresponding to the previous write position information.

Optionally, all bits in the initial page mask are 0.

Optionally, the target stripe satisfies the verification data reading condition when the number of write position information read reaches a preset number threshold, or the number of write position information read does not reach the preset number threshold, and the information record has no write position information to be read.

Optionally, converting the write position information into a page mask format to obtain a page mask corresponding to the write position information includes:

The initial page and page length corresponding to the write position information are calculated and converted into a page mask form to obtain a page mask corresponding to the write position information.

Optionally, recording the write position information of the target data block in each target stripe to the information record corresponding to the target stripe includes:

The write position information of each target data block in each target stripe is converted into a node, and the node is added to the linked list in the structure corresponding to the target stripe.

Optionally, reading the write position information from the information record in sequence includes: reading nodes from a linked list in sequence.

Optionally, the write position information is read from the information record in sequence, including:

The scheduling thread is used to read and write the position information from the information record in sequence.

Optionally, also include:

Based on the write position information of the target data block, old data in the target data block is read, and the old data and new data corresponding to the target data block are calculated to obtain an intermediate check value of the target data block.

Optionally, calculation is performed based on the old verification data to obtain new verification data, including:

Sequentially read the intermediate check value corresponding to each target data block, and after each intermediate check value of a target data block is read, perform an XOR operation on the intermediate check value and the first target check data to obtain the second target check data, until the intermediate check values of all target data blocks corresponding to the old check data are calculated, and the current second target check data is used as the new check data;

Among them, the first target verification data corresponding to the first intermediate verification value read is the old verification data, and the first target verification data corresponding to the intermediate verification value not read first is the second target verification data corresponding to the last intermediate verification value read.

Optionally, also include:

New data corresponding to each target data block is written into each target data block based on the write position information of each target data block.

Optionally, also include:

After the new data is written into the target data block, the new data is recorded into the buffer area.

The present application discloses a DRAID5 write processing device, comprising:

The stripe and information determination module is used to determine the target stripe to be written to in DRAID5 and the write position information of the target data block to be written to in each target stripe based on the IO write request initiated by the server; the write position information includes the write offset and the write length;

A stripe information recording module, used to record the write position information of the target data block in each target stripe to the information record corresponding to the target stripe;

A location information determination module, used to determine the full amount of location information of the check block in each target stripe that is currently to be updated based on the information record;

The old check data reading module is used to read data from the corresponding check block based on the current full position information when any target strip meets the check data reading condition, and obtain the old check data corresponding to the current full position information;

A new verification data calculation module, used to perform calculations based on old verification data to obtain new verification data;

The new verification data writing module is used to write the new verification data into the corresponding verification block based on the full position information.

The present application discloses an electronic device, including a memory and a processor, wherein:

Memory, used to store computer programs;

The processor is used to execute a computer program to implement the aforementioned disk array write processing method.

The present application discloses a computer non-volatile readable storage medium for storing a computer program, wherein the computer program implements the aforementioned disk array write processing method when executed by a processor.

It can be seen that the present application determines the target stripe to be written to data in DRAID5 and the write position information of the target data block to be written to data in each target stripe based on the IO write request initiated by the server; the write position information includes a write offset and a write length, and the write position information of the target data block in each target stripe is recorded in the information record corresponding to the target stripe, and the full position information of the check block in each target stripe to be updated is determined based on the information record. When any target stripe meets the check data reading condition, data is read from the corresponding check block based on the current full position information to obtain the old check data corresponding to the current full position information, and then calculation is performed based on the old check data to obtain new check data, and finally the new check data is written into the corresponding check block based on the full position information. That is, after the server initiates an IO write request, the present application records each data block to be written in each stripe to be written to the corresponding information record, and determines the full position information of the check block to be updated in each stripe based on the information record. When the check data reading condition is met, the check block is read once according to the current full position information, and all the old check data to be updated can be read. After calculation, the corresponding check block is written at one time based on the full position information, which reduces the number of check data reading and writing times, can reduce the time spent on processing check data, and thus improves storage performance.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

FIG1 is a schematic diagram of a block layout of a stripe in a disk array provided in an embodiment of the present application;

FIG2 is a flow chart of a disk array write processing method disclosed in an embodiment of the present application;

FIG3 is a schematic diagram of the structure of a disk array write processing device disclosed in an embodiment of the present application;

FIG. 4 is a schematic diagram of a structure of an electronic device disclosed in an embodiment of the present application.

Detailed ways

The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

The standard for distinguishing data blocks and check blocks in DRAID5 is stripnum (block code). The block code of the data block is a constant greater than 1 and less than the total number of blocks in the stripe. The block code of the check block is 0. See Figure 1. Figure 1 is a schematic diagram of the block layout of a stripe in a disk array, including disk 0, disk 1, disk 2, and disk 3. The stripe includes block 0 with a check code of 0, i.e., the check block, and blocks 1, block 2, and block 3 with codes of 1, 2, and 3, i.e., 3 data blocks. When a write IO process is initiated to the data block numbered 1 in disk 1, the offset and length in the block corresponding to the write IO are counted as stripoffset, count. The specific algorithm is as follows:

Step 1: To ensure the consistency of the stripe, it is necessary to recalculate the new check block of this stripe, read the old data within the stripoffset position count length of the data block of block number 1, and fetch (i.e., obtain from the server) the new data to be written within the stripoffset position count length of the data block of block number 1, and calculate the intermediate check value; Step 2: Write the new data to disk 1, the stripoffset position count length of the data block of block number 1; Step 3: Read the old check data within the stripoffset position count length of the check block in the strip of disk 0; Step 4: XOR the intermediate check value of the first step with the old check data read in the third step to obtain the new check data; Step 5: Start to write the new check data calculated in the fourth step to the disk and write it to the stripoffset position count length of the check block. Step 6: Io ends. It can be understood that when writing data to block codes 1 and 2, it is necessary to read the corresponding offset and length of the check block of the stripe respectively, and it is necessary to read the old check data twice, calculate the new check data twice, and write the new check data twice, which can be completed only once. The process of completing an IO is very complicated, and the problem that comes with it is that performance will also be affected. When the storage starts to initiate the write process, tens of thousands of IOs per second will bring a considerable time-consuming problem. To this end, this application provides a disk array write processing solution that can reduce the time consumption of processing verification data, thereby improving storage performance.

As shown in FIG. 2 , the embodiment of the present application discloses a disk array write processing method, including:

Step S11: Determine the target stripe to be written to in DRAID5 and the write position information of the target data block to be written to in each target stripe based on the IO write request initiated by the server; the write position information includes the write offset and the write length.

It is understandable that when the server initiates an IO write request to the storage, each IO write request will first be split into stripes and then into blocks to obtain the write offset and write length of each data block.

In a specific implementation, the old data in the target data block can be read based on the write position information of the target data block, and the old data and the new data corresponding to the target data block are calculated to obtain the intermediate check value of the target data block. Specifically, the old data corresponding to the write position information is XORed with the new data to obtain the intermediate check value corresponding to the target data block, and the intermediate check value is recorded.

Furthermore, the new data corresponding to each target data block can be written based on the write position information of each target data block. After writing the new data into the target data block, the new data is recorded in the cache area. When the next IO is initiated, if the data block is read, it can be determined whether it hits the cache. If so, the data is read from the cache without reading from the disk.

Step S12: Record the write position information of the target data block in each target stripe to the information record corresponding to the target stripe.

It can be understood that the write position information of the data block and the write position information of the corresponding check block in DRAID5 are the same.

In one embodiment, the information corresponding to each stripe is recorded as a structure; the write position information of each target data block in each target stripe is converted into a node, and the node is added to the linked list in the structure corresponding to the target stripe, waiting for thread scheduling.

Step S13: Determine the full amount of location information of the parity blocks in each target stripe that are currently to be updated based on the information record.

The page mask is in the form of a binary code with a preset number of bits, each bit corresponds to one page, and the target page mask corresponding to the first write position information read is the initial page mask, and the target page mask corresponding to the non-first write position information is the position information mask corresponding to the previous write position information. All bits of the initial page mask are 0.

Furthermore, the write position information is converted into a page mask format to obtain a page mask corresponding to the write position information. Specifically, the initial page and page length corresponding to the write position information are calculated, and converted into a page mask format to obtain a page mask corresponding to the write position information.

Taking the block size of 256kb as an example, the unit corresponding to the write offset and the write length is sector, and the conversion unit is 1kb=2sector, 1page=4kb. Take the write offset=32sector, and the write length=16sector as an example. 32sector=4page, 16sector=2page, a block of 256kb is converted into pages, which has 64 pages. The write offset is the fourth page, and the page length to be written is 2page. The corresponding page mask is 0 (58 zeros are omitted here) 11000.

A page mask variable may be used, the page mask variable records the initial page mask, each time the write position information is converted into a page mask, an OR operation is performed with the page mask variable, and the calculation result is updated to the page mask variable.

In one implementation, the write position information is read sequentially from the information record using a scheduling thread, and the nodes can be read sequentially from the linked list.

Step S14: When any target stripe meets the verification data reading condition, data is read from the corresponding verification block based on the current full position information to obtain old verification data corresponding to the current full position information.

The target stripe satisfies the verification data reading condition when the number of read write position information reaches a preset number threshold, or the number of read write position information does not reach the preset number threshold, and the information record has no write position information to be read. For example, when the number of read write position information does not reach 20, the write position information is no longer read.

That is, all the positions of the current check block where data update is to be performed can be determined, and all the corresponding old check data can be read at one time.

Step S15: Calculate based on the old verification data to obtain new verification data.

In one implementation, the intermediate check value corresponding to each target data block may be read in sequence, and after each intermediate check value of a target data block is read, the intermediate check value is XORed with the first target check data to obtain the second target check data, until the intermediate check values of all target data blocks corresponding to the old check data are calculated, and the current second target check data is used as the new check data;

For example, the full amount of position information corresponds to 20 pieces of write position information, and then 10 intermediate check values are used in sequence to perform an XOR operation.

Step S16: writing the new verification data into the corresponding verification block based on the full position information.

The disk array write processing solution provided by the present application is further described below. The algorithm can be implemented using a programming language. Regardless of whether the target data blocks are continuous, the merged reading and writing of the check blocks can be realized, including the following steps:

Step 1: Add member variables to the cache structure sce corresponding to the check block:
Struct sce{
Queue sce_queue;
Uint64PAGE_MASK;
}
Among them, queue is of the following type
Struct queue{
Struct queue*pre;
Struct queue*next;
}
Uint64PAGE_MASK is initialized to 0.

Each stripe is set with a structure sde, which stores pointer variables of the cache structure sce of all blocks (data blocks and check blocks) in the stripe. Cache structure function: When new data is written to the disk, the new data will be moved to the cache area and marked with the sce structure. If the next IO is initiated to read the block, the sce will query the mark to see if the cache hits. If the cache hits, the data does not need to be read from the disk.

Struct sde{

Sce*extension[16]

Note: The maximum number of blocks in a stripe is 16.

Step 2: Each time an IO is initiated, the IO will be split into stripes and then into blocks. For each block, a corresponding ipk (a structure that records the offset and length of the data block to be written) and a check block pipk (a structure that records the offset and length of the check block to be written) in the stripe corresponding to the block will be established. When the server initiates an IO to the storage, considering the case of multiple IOs, the write offset and write length of the data block are recorded as stripoffset i, count i, then the block offset and write length of the check block corresponding to the data block are stripoffset i, count i, and stripoffset i, count i are recorded in pipk i.

Step 3: Read the old data within the length of count i at position stripoffset i in the data block, fetch the new data to be written within the length of count i at position stripoffset i in the data block, and calculate the intermediate checksum paritybuilt i;

Step 4: Write new data to the data block stripoffset i position count i length in the stripe corresponding to the disk;

Step 5: Find the check block pipk i corresponding to the data block to which the new data is to be written, convert pipk i into a node form and add it to the node position in the Queue sce_queue linked list in the sce structure corresponding to the check block in sde, and wait for the thread scheduling to take out the task from the linked list.

Step 6: If there are other io, continue to execute the process of step 5. If other pipk j corresponds to the sce block of the same sde structure as pipk i in the fifth step, then convert this pipk j into a node form and add it to the next node position of the node in the Queue sce_queue linked list, still waiting for thread scheduling.

Step 7: The thread scheduling obtains the first node from the Queue sce_queue linked list.

Step 8: Convert the node to pipk i, and record its stripoffset i, count i. Convert stripoffset i, count i into the pagemask code corresponding to the page, perform an OR operation with PAGE_MASK, and update the calculation result to PAGE_MASK. Taking the block size of 256kb as an example, the unit corresponding to stripoffset and count is sector, and the conversion unit is 1kb=2sector. Converted to the form corresponding to page, 1page=4kb, take stripoffset i=32sector, count i=16sector as an example. 32sector=4page, 16sector=2page, a block of 256kb is converted into 64 pages, the offset to be written is the fourth page, and the page length to be written is 2page. The corresponding pagemask is 0 (58 zeros are omitted here) 11000.

Step 9: Determine whether there is a next node member in this linked list. If so, take out the next node, accumulate the number of merges coalescecount++, and limit the maximum value of the merged number coalescecountmax=20. If coalescecount=<coalescecountmax, execute step 8; if there is no other node after this node (that is, no other pipk is added to the sce->queue linked list during the period from pipk being added to the sce->queue linked list to the scheduling of taking out pipk from the sce->queue linked list, that is, if the next node is taken from the scequeue linked list and found to be empty, then no further waiting for merging is done) or the number of merges reaches coalescecountmax, execute step 10.

Step 10: Read the old verification data from the disk according to the value corresponding to the PAGEMASK code. After reading, execute step 11.

Step 11: XOR the old checksum data read in step 10 with the intermediate checksum value calculated in step 3 corresponding to pipk to obtain new checksum data. Go to step 12.

Step 12: Determine whether there are other nodes behind the node in the sce structure corresponding to this pipk. If not, it means that there is no verification merge. The new verification data calculated in step 11 is the final new verification data, and go to step 14. If there are other nodes behind the corresponding node and it has not reached 20, it means that there is still verification merge. The new verification data calculated in step 11 is not the final verification data, and go to step 13.

Step 13: Take the next node from the linked list and convert it into the corresponding pipk, find its corresponding intermediate check value, and perform XOR operation with the new check data obtained in step 11. After XOR is completed, continue to take the next node from the linked list and continue XOR operation until all nodes corresponding to the old check data in step 10 in the linked list are traversed, and the final new check data is calculated. Go to step 14.

Step 14: Start writing the calculated final new verification data to disk.

Step 15: Io is completed.

In this way, the problem of complex and time-consuming reading and writing of verification blocks in the write process can be solved, the number of read and write times initiated to the disk in IO can be reduced, the IOPS (Input/Output Operations Per Second) of the storage system can be increased, and storage performance can be effectively improved.

It can be seen that the embodiment of the present application determines the target stripe to be written to data in DRAID5 and the write position information of the target data block to be written to data in each target stripe based on the IO write request initiated by the server; the write position information includes a write offset and a write length, and the write position information of the target data block in each target stripe is recorded to the information record corresponding to the target stripe, and the full position information of the current data update to be performed on the check block in each target stripe is determined based on the information record. When any target stripe meets the check data reading condition, data is read from the corresponding check block based on the current full position information to obtain the old check data corresponding to the current full position information, and then calculation is performed based on the old check data to obtain new check data, and finally the new check data is written to the corresponding check block based on the full position information. That is, after the server initiates an IO write request, the embodiment of the present application records each data block to be written in each stripe to be written to the corresponding information record, and determines the full position information of the check block to be updated in each stripe based on the information record. When the check data reading condition is met, the check block is read once according to the current full position information, and all the old check data to be updated can be read. Then, calculations are performed and the corresponding check block is written at one time based on the full position information, which reduces the number of check data reading and writing times, can reduce the time spent on processing check data, and thus improves storage performance.

As shown in FIG3 , the embodiment of the present application discloses a DRAID5 write processing device, including:

The stripe and information determination module 11 is used to determine the target stripe to be written to in DRAID5 and the write position information of the target data block to be written to in each target stripe based on the IO write request initiated by the server; the write position information includes the write offset and the write length;

The stripe information recording module 12 is used to record the write position information of the target data block in each target stripe to the information record corresponding to the target stripe;

A location information determination module 13, used to determine the full location information of the check blocks in each target stripe that are currently to be updated based on the information record;

The old check data reading module 14 is used to read data from the corresponding check block based on the current full position information when any target strip meets the check data reading condition, so as to obtain the old check data corresponding to the current full position information;

A new verification data calculation module 15, used to perform calculation based on the old verification data to obtain new verification data;

The new verification data writing module 16 is used to write the new verification data into the corresponding verification block based on the full position information.

It can be seen that the embodiment of the present application determines the write position information of the target stripe to be written to and the target data block to be written to in each target stripe in DRAID5 based on the IO write request initiated by the server; the write position information includes a write offset and a write length, and the write position information of the target data block in each target stripe is recorded to the information record corresponding to the target stripe, and the full position information of the check block to be updated in each target stripe is determined based on the information record. When any target stripe meets the check data reading condition, a data read is performed from the corresponding check block based on the current full position information to obtain the old check data corresponding to the current full position information, and then the calculation is performed based on the old check data to obtain the new check data, and finally the new check data is written to the corresponding check block based on the full position information. That is, after the server initiates an IO write request, the embodiment of the present application records each data block to be written to in each stripe to be written to the corresponding information record, and determines the full position information of the check block to be updated in each stripe based on the information record. When the check data reading condition is met, the current full position information is read according to the current full position information. By reading the check block once, all the old check data to be updated can be read, and then calculations are performed, and the corresponding check block is written at one time based on the full location information, which reduces the number of times the check data is read and written, and can reduce the time spent on processing the check data, thereby improving storage performance.

In one embodiment, the position information determination module 13 is specifically used to read the write position information from the information record in sequence, and each time a write position information is read, the write position information is converted into a page mask form to obtain a page mask corresponding to the write position information, and the page mask is ORed with the target page mask to obtain a position information mask corresponding to the write position information, and the position information mask is recorded as the full position information of the parity block in the target stripe that is currently to be updated with data;

Furthermore, the target stripe satisfies the verification data reading condition when the number of the read write position information reaches a preset number threshold, or the number of the read write position information does not reach the preset number threshold, and the information record has no write position information to be read.

The step of converting the write position information into a page mask format to obtain a page mask corresponding to the write position information includes: calculating an initial page and a page length corresponding to the write position information, and converting the initial page and the page length into a page mask format to obtain a page mask corresponding to the write position information.

In one embodiment, the stripe information recording module 12 is specifically used to convert the write position information of each target data block in each target stripe into a node, and add the node to the linked list in the structure corresponding to the target stripe. Further, the position information determination module 13 is specifically used to read the nodes from the linked list in sequence.

In one implementation, the location information determination module 13 is specifically configured to sequentially read and write location information from information records using a scheduling thread.

Furthermore, the device also includes an intermediate check value calculation module, which is used to read old data in the target data block based on the write position information of the target data block, and calculate the old data and the new data corresponding to the target data block to obtain the intermediate check value of the target data block.

Correspondingly, the new verification data calculation module 15 is used to read the intermediate verification value corresponding to each target data block in turn, and after each intermediate verification value of a target data block is read, the intermediate verification value is XOR-operated with the first target verification data to obtain the second target verification data, until the intermediate verification values of all target data blocks corresponding to the old verification data are calculated, and the current second target verification data is used as the new verification data;

The first target verification data corresponding to the first intermediate verification value read is the old verification data, and the first target verification data corresponding to the first intermediate verification value read is the old verification data. The first target verification data corresponding to the intermediate verification value is the second target verification data corresponding to the last read intermediate verification value.

Furthermore, the device further comprises a data block writing module, which is used to write new data corresponding to each target data block into each target data block based on the writing position information of each target data block.

In addition, the device also includes a write cache module, which is used to record the new data into the cache area after writing the new data into the target data block.

As shown in FIG. 4 , an embodiment of the present application discloses an electronic device 20, including a processor 21 and a memory 22; wherein the memory 22 is used to store a computer program; and the processor 21 is used to execute the computer program to implement the following steps:

Based on the IO write request initiated by the server, the write position information of the target stripe to be written to data in DRAID5 and the target data block to be written to data in each target stripe is determined; the write position information includes the write offset and the write length; the write position information of the target data block in each target stripe is recorded to the information record corresponding to the target stripe respectively; based on the information record, the full position information of the check block in each target stripe to be updated with data is determined; when any target stripe meets the check data reading condition, data is read from the corresponding check block based on the current full position information to obtain the old check data corresponding to the current full position information; calculation is performed based on the old check data to obtain new check data; and the new check data is written into the corresponding check block based on the full position information.

In this embodiment, when the processor 21 executes the computer subroutine stored in the memory 22, the following steps can be specifically implemented: Step: read the write position information from the information record in sequence, and each time a write position information is read, convert the write position information into a page mask form to obtain a page mask corresponding to the write position information, and perform an OR operation on the page mask and the target page mask to obtain a position information mask corresponding to the write position information, and record the position information mask as the full amount of position information of the parity block in the target stripe that is currently to be updated;

In this embodiment, when the processor 21 executes the computer subroutine stored in the memory 22, the following steps can be specifically implemented: calculating the initial page and page length corresponding to the write position information, and converting them into a page mask form to obtain a page mask corresponding to the write position information.

In this embodiment, when the processor 21 executes the computer subroutine stored in the memory 22, the following steps can be specifically implemented: converting the write position information of each target data block in each target stripe into a node, and adding the node to the linked list in the structure corresponding to the target stripe.

In this embodiment, when the processor 21 executes the computer subroutine stored in the memory 22, the following steps may be specifically implemented: reading nodes from the linked list in sequence.

In this embodiment, when the processor 21 executes the computer subroutine stored in the memory 22, the following steps may be specifically implemented: using the scheduling thread to sequentially read the write position information from the information record.

In this embodiment, when the processor 21 executes the computer subroutine stored in the memory 22, the following steps can be specifically implemented: based on the write position information of the target data block, the old data in the target data block is read, and the old data and the new data corresponding to the target data block are calculated to obtain the intermediate check value of the target data block.

In this embodiment, when the processor 21 executes the computer subroutine stored in the memory 22, the following steps can be specifically implemented: sequentially reading the intermediate check value corresponding to each target data block, and after each intermediate check value of a target data block is read, performing an XOR operation on the intermediate check value and the first target check data to obtain the second target check data, until the intermediate check values of all target data blocks corresponding to the old check data are calculated, and the current second target check data is used as the new check data;

The first target verification data corresponding to the first intermediate verification value read is the old verification data, and the first target verification data corresponding to the intermediate verification value not read first is the second target verification data corresponding to the last intermediate verification value read. according to.

In this embodiment, when the processor 21 executes the computer subroutine stored in the memory 22, the following steps may be specifically implemented: based on the write position information of each target data block, new data corresponding to each target data block is written into each target data block.

In this embodiment, when the processor 21 executes the computer subroutine stored in the memory 22, the following steps may be specifically implemented: after writing the new data into the target data block, the new data is recorded into the cache area.

Furthermore, the memory 22, as a carrier for storing resources, may be a read-only memory, a random access memory, a magnetic disk or an optical disk, etc., and the storage method may be temporary storage or permanent storage.

In addition, the electronic device 20 also includes a power supply 23, a communication interface 24, an input/output interface 25 and a communication bus 26; wherein the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input/output interface 25 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

Furthermore, an embodiment of the present application discloses a computer non-volatile readable storage medium for storing a computer program, wherein when the computer program is executed by a processor, the following steps are implemented:

Based on the IO write request initiated by the server, the write position information of the target stripe to be written to data in DRAID5 and the target data block to be written to data in each target stripe is determined; the write position information includes the write offset and the write length; the write position information of the target data block in each target stripe is recorded to the information record corresponding to the target stripe respectively; based on the information record, the full position information of the check block in each target stripe to be updated with data is determined; when any target stripe meets the check data reading condition, data is read from the corresponding check block based on the current full position information to obtain the old check data corresponding to the current full position information; calculation is performed based on the old check data to obtain new check data; based on the full position information, the new check data is written into the corresponding check block.

In this embodiment, when the computer subroutine stored in the non-volatile computer readable storage medium is executed by the processor, the following steps can be specifically implemented: read the write position information from the information record in sequence, and each time a write position information is read, convert the write position information into a page mask form to obtain a page mask corresponding to the write position information, perform an OR operation on the page mask and the target page mask to obtain a position information mask corresponding to the write position information, and record the position information mask as the full amount of position information of the parity block in the target stripe that is currently to be updated;

In this embodiment, when the computer subroutine stored in the computer non-volatile readable storage medium is executed by the processor, the following steps can be specifically implemented: calculating the initial page and page length corresponding to the write position information, and converting them into a page mask form to obtain the page mask corresponding to the write position information.

In this embodiment, when the computer subroutine stored in the computer non-volatile readable storage medium is executed by the processor, the following steps can be specifically implemented: converting the write position information of each target data block in each target stripe into a node, and adding the node to the linked list in the structure corresponding to the target stripe.

In this embodiment, when the computer subroutine stored in the computer non-volatile readable storage medium is executed by the processor, the following steps may be specifically implemented: reading nodes from the linked list in sequence.

In this embodiment, when the computer subroutine stored in the computer non-volatile readable storage medium is executed by the processor, the following steps can be specifically implemented: using the scheduling thread to sequentially read the write position information from the information record.

In this embodiment, when the computer subroutine stored in the computer non-volatile readable storage medium is executed by the processor, the following steps can be specifically implemented: based on the write position information of the target data block, the old data in the target data block is read, and the old data and the new data corresponding to the target data block are calculated to obtain the intermediate check value of the target data block.

In this embodiment, when the computer subroutine stored in the non-volatile computer readable storage medium is executed by the processor, the following steps can be specifically implemented: sequentially reading the intermediate check value corresponding to each target data block, and after each intermediate check value of a target data block is read, performing an XOR operation on the intermediate check value and the first target check data to obtain the second target check data, until the intermediate check values of all target data blocks corresponding to the old check data are calculated, and the current second target check data is used as the new check data;

In this embodiment, when the computer subroutine stored in the computer non-volatile readable storage medium is executed by the processor, the following steps can be specifically implemented: writing new data corresponding to each target data block into each target data block based on the write position information of each target data block.

In this embodiment, when the computer subroutine stored in the computer non-volatile readable storage medium is executed by the processor, the following steps can be specifically implemented: after writing the new data into the target data block, the new data is recorded into the cache area.

In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

The above is a detailed introduction to a disk array write processing method, device, equipment and medium provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of the present application. At the same time, for a person skilled in the art, according to the idea of the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

A disk array write processing method, characterized by comprising:

Determine the target stripe to be written to data in DRAID5 and the write position information of the target data block to be written to data in each target stripe based on the IO write request initiated by the server; the write position information includes the write offset and the write length;

Recording the write position information of the target data block in each target stripe to the information record corresponding to the target stripe;

Determine the full amount of location information of the parity block in each target stripe that is currently to be updated based on the information record;

When any of the target stripes meets the check data reading condition, data is read from the corresponding check block based on the current full position information to obtain old check data corresponding to the current full position information;

Perform calculation based on the old verification data to obtain new verification data;

The new verification data is written into the corresponding verification block based on the full position information.
The disk array write processing method according to claim 1, characterized in that the step of determining the full amount of location information of the parity block in each target stripe that is currently to be updated with data based on the information record comprises:

The write position information is read from the information record in sequence, and each time a write position information is read, the write position information is converted into a page mask form to obtain a page mask corresponding to the write position information, and the page mask is ORed with the target page mask to obtain a position information mask corresponding to the write position information, and the position information mask is recorded as the full amount of position information of the parity block in the target stripe that is currently to be updated with data;

Among them, the page mask format is a binary code format of a preset number of bits, each bit corresponds to a page, and the target page mask corresponding to the first write position information read is the initial page mask, and the target page mask corresponding to the non-first write position information is the position information mask corresponding to the previous write position information.
The disk array write processing method according to claim 2, characterized in that each bit of the initial page mask is 0.
The disk array write processing method according to claim 2 is characterized in that the target stripe satisfies the verification data reading condition when the number of write position information read reaches a preset number threshold, or the number of write position information read does not reach the preset number threshold, and the information record has no write position information to be read.
The disk array write processing method according to claim 2, characterized in that the converting the write position information into a page mask form to obtain a page mask corresponding to the write position information comprises:

The initial page and page length corresponding to the write position information are calculated and converted into a page mask form to obtain a page mask corresponding to the write position information.
The disk array write processing method according to claim 2, characterized in that the step of recording the write position information of the target data block in each target stripe to the information record corresponding to the target stripe comprises:

The write position information of each target data block in each target stripe is converted into a node, and the node is added to the linked list in the structure corresponding to the target stripe.
The disk array write processing method according to claim 6, characterized in that the sequentially reading the write position information from the information record comprises: sequentially reading nodes from the linked list.
The disk array write processing method according to claim 2, wherein the step of sequentially reading the write position information from the information records comprises:

The scheduling thread is used to read and write the position information from the information record in sequence.
The disk array write processing method according to any one of claims 1 to 8, characterized in that it also includes:

Based on the write position information of the target data block, old data in the target data block is read, and the old data and new data corresponding to the target data block are calculated to obtain an intermediate check value of the target data block.
The disk array write processing method according to claim 9, wherein the step of calculating based on the old check data to obtain the new check data comprises:

Sequentially read the intermediate check value corresponding to each target data block, and after each intermediate check value of a target data block is read, perform an XOR operation on the intermediate check value and the first target check data to obtain the second target check data, until the intermediate check values of all the target data blocks corresponding to the old check data are calculated, and the current second target check data is used as the new check data;

Among them, the first target verification data corresponding to the first intermediate verification value read is the old verification data, and the first target verification data corresponding to the intermediate verification value not read first is the second target verification data corresponding to the last intermediate verification value read.
The disk array write processing method according to claim 9, characterized in that it also includes:

The new data corresponding to each of the target data blocks is written into each of the target data blocks based on the write position information of each of the target data blocks.
The disk array write processing method according to claim 11, characterized in that it also includes:

After the new data is written into the target data block, the new data is recorded into the buffer area.
The disk array write processing method according to claim 12, characterized in that it also includes:

When a new IO is initiated to read the target data block, it is determined whether the data block hits the cache.

In response to hitting the cache, data is read from the cache.
The disk array write processing method according to claim 5, characterized in that it also includes:

The initial page mask is recorded into a page mask variable.
The disk array write processing method according to claim 14, characterized in that it also includes:

An OR operation is performed on the page mask and the initial page mask, and the operation result is updated to the page mask variable.
The disk array write processing method according to claim 9, characterized in that it also includes:

The intermediate check value is recorded.
The disk array write processing method according to claim 11, characterized in that the write position information of the target data block in the target stripe is the same as the write position information of the corresponding check block in the target stripe.
A DRAID5 write processing device, comprising:

A stripe and information determination module, used to determine the target stripe to be written to in DRAID5 and the write position information of the target data block to be written to in each target stripe based on the IO write request initiated by the server; the write position information includes a write offset and a write length;

A stripe information recording module, used for recording the write position information of the target data block in each target stripe to the information record corresponding to the target stripe;

A location information determination module, used to determine the full amount of location information of the check block in each target stripe that is currently to be updated based on the information record;

An old check data reading module, used for reading data from the corresponding check block based on the current full position information when any of the target stripes meets the check data reading condition, to obtain the old check data corresponding to the current full position information;

A new verification data calculation module, used to perform calculation based on the old verification data to obtain new verification data;

A new verification data writing module is used to write the new verification data into the corresponding verification block based on the full position information.
An electronic device, comprising a memory and a processor, wherein:

The memory is used to store the computer program;

The processor is used to execute the computer program to implement the disk array write processing method according to any one of claims 1 to 17.
A computer non-volatile readable storage medium, characterized in that it is used to store a computer program, wherein the computer program, when executed by a processor, implements the disk array write processing method according to any one of claims 1 to 17 Management method.