JPH07306758A - Disk array device and its control method - Google Patents

Abstract

PURPOSE:To alter the constitution of a data file and to improve the reliability or performance of the whole device by making the redundancy of data variable during an on-line connection with a host device or turning OFF the power source of this device. CONSTITUTION:An MPU 240 executes commands in a RAM 250 while decoding them sequentially to control the whole disk drive controller 200. At a user's request to vary the reliability of an optical logical data file, the MPU 240 of a disk drive controller 200 sets a flag (in-redundancy-variation information) indicating that the redundancy of redundancy variation information in the RAM 250 is being varied and requested redundancy (variation requested redundancy). Therefore, the number of redundant data blocks can be varied even while the disk array device is connected to the host on an on-line basis or without stopping the disk array device; and the redundancy of the logical data file is varied to flexibly comply with a request for the reliability of user data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array device for distributing and storing a logical data file into a plurality of data, and more particularly to a method of changing a redundant data storage drive of the device (disk array subsystem).

[0002]

2. Description of the Related Art A conventional device is disclosed in Japanese Patent Laid-Open No. 62-24481.
As described in Japanese Patent Publication No. JP-A-2003-187, the predetermined performance is realized by dividing the logical data from the host device into a plurality of data blocks and accessing a plurality of storage media in parallel. A parity data block is created for each unit accessed in parallel, and the data block is written in parallel to the parity data storage dedicated area of the storage medium at the same time as the writing of data. And when reading data,
When a read error occurs, the data blocks other than the faulty data block and the parity data block are read out, the faulty data block is restored from these data blocks, and transferred to the host device to maintain high reliability. ing.

[0003]

In the prior art, there is a problem that the system is fixed and it is not possible to meet various requests of the user. That is, in the conventional technology, once a file system is introduced, this system has the same reliability. When higher reliability was required, it was necessary to stop the entire system or a part of the system (power off) and replace hardware or software.

For example, a certain logical data file is assigned to a disk array device with a redundancy of 1 and a division number of 4, and a system is introduced. That is, any logical data in the logical data file is divided into four physical data blocks and is stored in the storage medium together with one parity block. In this way, when a failure occurs in any one physical data block in the four physical data blocks in the access to this logical data file, the data block and parity block other than the failed data block are read out, and the parity generation circuit is read. The logical data can be restored by restoring the failed data block. However, since this RAID configuration has a redundancy of 1, data cannot be restored if two or more physical data blocks among the four physical data blocks constituting any logical data fail. Here, RAID means Redundant Arrays o
This is a configuration of a disk array device called fInexpensive Disks and is classified into Class 1 to Class 5 for convenience.

When there is a request from the user to further improve the reliability of this logical data file, this RAI
In the D configuration, for example, it is necessary to increase the redundancy to 2. At this time, in the conventional technology, it is impossible to change the configuration of the logical data file while the disk array device and the host are online, so the disk array device is temporarily stopped, the power is turned off, and then the power is turned on again. Then I had to reintroduce the system.

As another example, when the logical data file has a RAID class 5 configuration in which the data distribution unit is a plurality of logical data files, a write request from a higher-level device (host) causes an update target. It is necessary to read the data block and the parity data block from the drive once, generate a new parity data block from the old data block, the old parity data block, and the update data block, and write the update data block and the new parity block. In other words, the reliability can be maintained because the failure block can be restored for one failure of the data block, but the above processing is required at the time of a write request, and the performance is degraded (this phenomenon is referred to as "write penalty"). Call).

When a user requests that the logical data file requires performance rather than reliability, it is necessary to change the RAID configuration to, for example, redundancy level 0. Since it was impossible to change the configuration of the data file, the disk array device had to be stopped and the system installed again.

An object of the present invention is to make it possible to change the redundancy level of a disk array device while it is online connected to a host device or without shutting off the power supply of the disk array device. It is an object of the present invention to provide a disk array device capable of changing the reliability of the device and the performance thereof. Another object of the present invention is to enable I / O processing from a higher-level device for a logical data file while changing the redundancy, and prevent performance degradation of the disk array device due to changing the redundancy. Another object of the present invention is to enable the device to be stopped (power cutoff) even while the disk array device is being changed, and to operate the disk array device when the device is restarted. For the purpose of improvement, the change processing of the disk array device is continuously executed without the intervention of maintenance personnel.

[0009]

The main constitution of the present invention is as follows: 1) a parity generation circuit for generating parity data of one or more redundancy levels; 2) temporarily storing write request data from a host; or A data buffer that temporarily stores a data block / parity block from a physical drive and stores one or more parity blocks generated by the parity generation circuit, 3) manages the RAID configuration of a logical data file specified by the user RAID management information; this information includes the following: a) number of divisions / distribution unit / redundancy of each logical data file, b) rotation unit of redundant data, and c) each logical data file is stored. Physical drive address, 4) redundancy change information requested by the user; this information includes: d) Redundancy change in progress information, change request redundancy, change end pointer, e) storage target physical drive information of logical data file after redundancy change, 5) physical drive information; this information includes the following: , F) logical data file number stored in the physical drive, g) physical drive data identifier (data / parity /
Preliminary), 6) Non-volatile memory for storing the information of 3), 4), 5) above, 7) Storage address of logical data (hereinafter, the address is AD
R)) and the RAID management information in 3) above,
Storage target physical drive ADR, physical drive data AD
An address translation mechanism for calculating R is included.

[0010]

When the user makes a request for changing the reliability of an arbitrary logical data file, the microprocessor (described later) of the disk drive controller is changing the redundancy of the redundancy changing information in the non-volatile memory. A flag indicating that (redundancy changing information) and the redundancy requested by the user (change request redundancy) are set.

If the user's request is a redundancy reduction request, the rotation unit of the RAID management information is referred to,
If it is not rotated, the redundancy of the RAID management information and the physical drive ADR are changed, and the processing is ended. Here, the rotate refers to an operation of storing redundant data by exchanging (turning) in a certain unit for all physical drives, instead of fixedly storing the redundant data in one certain physical drive.

When rotating, the redundancy changing information of the redundancy changing information, the requested redundancy number and the changed physical drive ADR information are set. Then, the data block and parity block are read from the physical drive where the file is stored into the data buffer, the storage position in the physical drive is calculated using the address conversion mechanism, excluding the parity block for the redundancy reduction, and the redundancy is calculated. The data block and the parity block are written to the physical drive excluding the parity block for the reduction.

At the end of this writing, the change end pointer of the redundancy change information is updated. These processes are repeated until all the logical data files are completed. After the completion, the redundancy of the RAID management information and the physical drive ADR are changed, and the redundancy changing information of the redundancy changing information is reset.

When the user's request is a request to increase the redundancy, the redundancy changing information of the redundancy changing information, the requested redundancy number, and the changed physical drive ADR information are set.

When not rotated, only the data block is read from the physical drive storing the file into the data buffer, and the parity generation circuit generates the increased parity block. Then, the increased parity block is written to the storage target physical drive. At the end of writing, the change end pointer of the redundancy change information is updated. These processes are repeated until all logical data files are completed.

When rotated, the data block and the parity block are read from the physical drive in which the file is stored into the data buffer, and the parity generation circuit generates the increased number of parity blocks. A parity block corresponding to an increase in redundancy is added from the data block, the storage position in the physical drive is calculated using the address conversion mechanism, and the data block and the parity block are written in the storage target physical drive. At the end of writing,
The change end pointer of the redundancy change information is updated. These processes are repeated until all logical data files are completed.

When all the above changing processes are completed,
The redundancy of the RAID management information and the physical drive ADR storing the logical data file are changed, and the redundancy changing information of the redundancy changing information is reset. In accordance with the redundancy reduction request, the physical drive vacant due to the redundancy reduction is made available as a spare drive by the user's designation. At this time, the data identifier of the physical drive information is switched from parity to spare drive.

Next, a case where an access is made from the host during the processing of changing the redundancy will be described. ◆ During the above change process, if there is a read request from the host,
The microprocessor of the disk drive controller refers to the redundancy changing information of the redundancy change information of the read target logical data file, and refers to the change end pointer because it is being changed, and the read target logical data has already changed the redundancy. Ask if it is done or not.

If not changed, the data is read from the target drive into the data buffer and transferred to the host according to the RAID management information. ◆ If changed, according to the redundancy change information, if a physical drive is being added, read the data buffer including parity data and transfer only the data block to the host.

When a write request is issued from the host during the above change processing, the microprocessor of the disk drive controller refers to the redundancy changing information of the redundancy changing information of the write-targeted logical data file, and changes the information. Since it is in the middle, the change end pointer is referred to determine whether or not the logical data to be written has already been changed in redundancy. ◆ If not changed, the data is written from the data buffer to the physical drive in the target drive according to the RAID management information. ◆
If the data has been changed, the data including the parity data being added to the physical drive is written from the data buffer to the physical drive according to the redundancy change information.

As described above, even if the disk array device is online with the host, or the number of redundant data blocks can be changed without stopping the disk array device, the redundancy of the logical data file can be changed. By changing it, it is possible to flexibly meet the request for the reliability of the user data. ◆ In addition, when the data distribution is rotated, the data storage location changes due to the change in redundancy. At this time, in order to improve the reliability of the data, the data read during the change is stored in the data buffer, and after the redundancy is changed, the data whose storage position is changed is read again from the physical drive and the By comparing with the previous data in the buffer, it is possible to prevent the operation of writing the data by mistake due to the change. ◆ Also, it takes a long time to process the changes,
Since the redundancy change information is stored in the non-volatile memory, the subsystem can be stopped and the power can be turned off (P / S OFF) without waiting for the end of the processing. The next time the power is turned on, IMPL (Initial Micro P
After the program load, the redundancy change information is referenced first, and if the change is in progress, the redundancy change process is restarted from the end of the change end pointer, so that it can be automatically re-executed without the intervention of maintenance personnel. The operational performance of the disk array device is improved.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. 1 is a computer system including a disk drive control device to which the present invention is applied. This computer system includes a CPU 100 that is a central processing unit, a disk drive control device 200, and a disk drive device 300.
Mainly consists of and. Then, the disk drive control device 200 controls the disk drive device 300 according to an instruction from the CPU 100.

In addition, the status is displayed on the operation panel 201 as appropriate based on the redundancy changing information, the data identifier of the physical drive information, and other internal information of the disk drive control device 200. As a result, the system 1) the redundancy is being changed, 2) the physical drive attribute has switched the drive from the parity drive to the spare drive, and vice versa. 3) the host to the disk drive. It is clearly indicated to the operator that the control device 200 was accessed, 4) that the operation of changing the redundancy of the system was restarted by turning the power on again.

FIG. 2 shows a disk drive controller 200.
The internal structure of is shown. A microprocessor unit (hereinafter referred to as MPU) 240 executes a random access memory (hereinafter referred to as RAM) 250 while sequentially decoding, and controls the entire disk drive control device 200. The drive circuit 290 drives the operation panel 201 and also has an output terminal to another display device (not shown).

The channel controller 210 has a CPU 100.
And control data transfer with. Drive controller 22
0 controls data transfer with each drive. The data buffer 230 is a memory used when data is transferred between the channel controller 210 and the drive controller 220. This memory may be a volatile memory or a non-volatile memory. Here, a volatile memory will be described as an example.

The parity generation circuit 260 includes the CPU 100.
It has a function to generate redundant data for the data sent from it, and this function can also be used for data restoration. The unit for adding the redundant data may be one logical data unit sent from the host device or a plurality of logical data units. Also, multiple logical data are
It may be logical data in one data file viewed from the CPU 100, or logical data of a plurality of data files. Here, in the method of adding redundant data to four logical data and distributing and storing the data in six physical drives, the redundancy changing process will be described as an example.

The non-volatile memory 270 is set with information for performing the redundancy changing process for an arbitrary logical data file while the MPU 240 is online. By setting this information in the non-volatile memory 270, the power can be cut off (P / S OFF) even when the redundancy is changed, and the change processing can be automatically continued at restarting. There are two characteristics.

The address translation mechanism 280 is the CPU 100.
It is for calculating the physical drive number and the physical data address when accessing the disk drive device 300 from the logical data file number and the logical data address instructed by the RAID configuration of the logical data file.

FIG. 3 shows a plurality of magnetic disk drives constituting the disk drive device 300. The data transfer control devices 310 to 360 are for performing data transfer with the disk drive control device 200.
Each of the data transfer control devices 310 to 360 has four
Physical drives 310a-310d, 320a-32
0d, ..., 360a to 360d are connected.

In the present embodiment, four logical data files (Ea, Eb, Ec, Ed) are assigned to these physical drive groups.
Is set. In each logical drive group, Ea = drives 310a to 360a, Eb = 310
b to 360b, Ec = 310c to 360c, Ed = 31
0d to 360d, and the data recovery group has the same configuration.

Next, the table group in this embodiment is shown in FIG.
Starting from FIG. ◆ The RAID management information 400 shown in FIG. 4 corresponds to each logical data file Ea, Eb, Ec, E.
The number of divisions 401 indicating how many physical drives the logical data file divides for each d, and the distribution unit 40 indicating the unit of data stored in one physical drive.
2. Redundancy degree 403 indicating how much redundant data is set in the logical data file, whether redundant data is fixedly stored in one physical drive for the storage target physical drive group, or predetermined for all physical drives Rotate unit 404 that indicates whether to store (rotate) while rotating in units of A, and physical drive A that indicates the ADR (address) of the physical drive that stores the logical data file.
It is composed of the DR 405.

When the rotation unit 404 is 0, there is no rotation, that is, redundant data is stored in one fixed physical drive. When the rotate unit 404 is other than 0, the disk array device configures the RAID class 5 and distributes logical data and redundant data to those physical drives while rotating in units specified by the rotate unit 404.

The change management table 500 of FIG. 5 shows a request for changing the redundancy of a logical data file. The fact that the FLAG 501 being executed has a predetermined content indicates that the logical data file is undergoing the redundancy changing process. The redundancy count 502 indicates the redundancy count after the change. The storage end pointer 503 indicates the ADR at which the redundancy change of the logical data file is completed. The changed physical drive ADR information 504 is information indicating the changed physical drive group. ◆ When reducing the redundancy, one set of physical drives forming the corresponding logical data file will be left. To increase the redundancy, one set of physical drive groups will be added. ADR information 5
04 is the physical drive A including the addition of one set
It will indicate the DR information.

The physical drive information 600 in FIG. 6 indicates which logical data file is stored in each physical drive and the attribute of that data. ◆ The logical data file number 601 indicates the logical data file number stored in the physical drive. The identifier 602 indicates whether the physical drive is a drive to which logical data has been allocated or a spare drive. The RAID management information 40
0, change management table 500, physical drive information 600
Is set in the non-volatile memory 270 (FIG. 2).

FIG. 7 shows the data format used in this embodiment. ◆ One ECC group (parity generation unit) 70
0 consists of four logical data blocks (hereinafter referred to as data blocks) 701, 702, 703 and 704, and further redundant data blocks (hereinafter referred to as parity blocks).
705 is added by the parity generation circuit 260 (FIG. 2) (261; 1 parity generation process). ◆ Further, if the redundancy is increased from 1 to 2, the data blocks 701, 70
Two redundant data blocks 705 and 706 are created from 2, 703 and 704 (262; 2 parity generation processing). The five or six data blocks are, for example, physical drives 310a, 320a, 330a, 340 of FIG. 3 for physical drives connected in parallel.
Data is transferred to a, 350a, and 360a.

FIG. 8 shows the configuration of the disk array device RAID class 5 with rotation, and the RA of 801 is shown.
The conceptual diagram at the time of increasing one degree of redundancy from ID structure to 802 is shown. RAI of this logical data file 801
The D configuration has a division number 401 = 4, a distribution unit 402 = 2, a redundancy number 403 = 1, a rotation unit 404 = 2, and a physical drive ADR 405 = 310a / 320a / 330a /.
340a / 350a. Increase the redundancy by 1,
Redundancy 502 = 2, changed physical drive ADR information 504 = 310a / 320a / 330a / 340a / 3
It is 50a / 360a (802).

Next, the normal operation of the disk array device according to the present invention will be described with reference to FIG. When receiving a write request from the CPU 100, the disk drive control device 200 normally receives the write logical data by the channel control device 210 and temporarily stores it in the data buffer 230. The MPU 240 generates a parity block in the parity generation circuit 260 according to the redundancy degree 403 of the RAID management information 400 (FIG. 4) in the non-volatile memory 270. This parity block is temporarily stored in the data buffer 230. The MPU 240 has a division number 401, a distribution unit 402, and a redundancy degree 4 of the RAID management information 400.
03, rotate unit 404, physical drive ADR40
5 is input to the address translation mechanism 280. ◆ As a result, the MPU 240 recognizes the physical drive to which logical data is written and its ADR. According to the output ADR information, the MPU 240 writes the data block and the parity block in the physical drive and ends the write request from the CPU 100.

When there is a read request from the CPU 100, the MPU 240 determines that the RAID in the non-volatile memory 270 is
Number of divisions 401 of management information 400, distribution unit 402, redundancy degree 403, rotation unit 404, physical drive AD
By inputting R405 to the address conversion mechanism 280, the physical drive to which logical data is written and its AD
Recognize R. According to the output ADR information, MPU2
40 reads the data block / parity block into the physical drive and transfers it to the host via the data buffer 230. When there is a faulty block, the MPU 240 recovers the faulty block from the remaining blocks by the parity generation circuit 260 and transfers it to the host.

Next, FIG. 9 shows a method of controlling the disk array device when the user issues a redundant drive reduction instruction.
In the case where an expansion instruction is issued, it will be described with reference to the flowcharts shown in FIGS. ◆ In the reduction processing, the logical data file Ea (FIG. 3) is used in this embodiment.
The description will be made by taking the reduction from 2 parity to 1 parity as an example, but the change from 2 parity or 1 parity to no parity is also possible by the same processing method.

First, in STEP 901 (FIG. 9), RAI
D Refer to the rotation unit 404 of the management information 400, and E
It is determined whether or not a is a rotate. When there is a rotate, the process proceeds to STEP902 and the change management table 5
The running FLAG 501 for Ea of 00 is set, and the redundancy number 502 is set to the redundancy request = 1.
Further, the changed physical drive ADR information 504 includes the physical drive ADR 310a / 32 that stores the reduced Ea.
0a / 330a / 340a / 350a is set. ◆ Next, in STEP 903, the data block / parity block is read from the stored physical drive group. Therefore, the physical drive ADR40 of the RAID management information 400 is read.
The target physical drive ADR is obtained from 5. STEP90
4. Physical drive group 310a obtained in STEP 903
/ 320a / 330a / 340a / 350a / 360a
Then, the data block / parity block is read and stored in the data buffer 230 (RD).

At STEP 905, in order to find the physical drive group to store the changed logical data file,
Physical drive ADR after change in the change management table 500
The physical drive ADR is referred to from the information 504. That A
The DR information, the redundancy degree 502, the division number 401 of the RAID management information 400, and the distribution unit 402 are assigned to the address conversion mechanism 28.
By inputting 0, the storage ADR after redundancy reduction is calculated. ◆ Next, write 4 data blocks / 1 parity block to the physical drive in the ADR obtained in STEP 905 (WR: STEP 906).

When the WR is completed, the redundancy change information 5
The storage end pointer 503 of 00 is updated (STEP 9
07).

The processing from STEP 904 to STEP 907 is repeated until the processing is completed for all logical data files Ea. When it is determined in STEP 908 that the processing is completed, the process proceeds to STEP 909. ◆ The redundancy degree 403 of the RAID management information 400 is changed to the redundancy degree 502 of the redundancy change information 500, and the physical drive ADR 405 of the RAID management information 400 is changed to the changed physical drive ADR information 504 of the redundancy change information 500. ◆ STE
FLAG50 during execution of the redundancy change information 500 in P910
1 is reset, and the reduction process ends.

Next, when it is determined in STEP 901 that there is no rotation, the process proceeds to STEP 913, the redundancy degree 403 of the RAID management information 400 is changed to the requested number, and the RA
The physical drive ADR 405 of the ID management information 400 is replaced with the physical drive ADR information 5 after the change of the redundancy change information 500.
Change to 04. ◆ In STEP 911, it is determined whether or not the empty physical drive is used as a spare drive. When the empty physical drive is used as a spare drive, the process proceeds to STEP 912, and the data identifier 60 of the physical drive ADR in the physical drive information 600.
Set to 2 to use as a spare drive. In this way, the user can specify the usage method of the physical drive that has become free due to the reduction in redundancy.

When many failures occur in accessing a certain physical drive, the spare drive is searched by referring to the identifier 602 of the physical drive information 600. When a spare drive exists, the data of the failed drive is copied to the spare drive, the logical data file number 601 of the physical drive information 600 is copied from the failed drive to the spare drive, and the identifier 602 is also copied from the failed drive to the spare drive. . Further, the physical drive ADR 405 of the RAID management information 400 of the logical data file stored in the failed drive also switches from the failed drive ADR to the spare drive ADR. ◆ By controlling in this way,
It is also possible to use the reduced physical drive as a spare drive while online later. Further, the reduced physical drive may be assigned to store a new logical data file.

The feature of this embodiment is that the RAID management information 40
0, change management table 500 and physical drive information 60
By having a parity generation circuit 260 that has control information of 0 and can generate a plurality of parity blocks of a redundant number, STEP 904 to STEP 907, or STEP
By performing the processing of 913, the number of redundant logical data files can be reduced while the disk array apparatus and the host are online.

Next, a method of controlling the disk array device when adding physical drives will be described with reference to the flowchart of FIG. In the extension processing, in the present embodiment, an explanation will be given by taking an example of extension from 1 parity to 2 parity for the logical data file Ea, but a change from no parity to 1 parity or 2 parity can be made by the same processing method.

First, in STEP 1001, the running FLAG 501 for Ea of the change management table 500 is set, and the redundancy number 502 is set to the redundancy request = 2. Further, the changed physical drive ADR information 504 includes the physical drive ADR 310 that stores the added Ea.
a / 320a / 330a / 340a / 350a / 360
Set a. Next, in STEP 1002, a data block / parity block is read from the stored physical drive group, so the target physical drive ADR is obtained from the physical drive ADR 405 of the RAID management information 400. ◆ In STEP1003, physical drive groups 310a / 320a / 330a / 34 obtained in STEP1002
The data block / parity block is read from 0a / 350a and stored in the data buffer 230 (FIG. 2) (RD).

Next, two parities are generated by the parity generation circuit 260 and the additional parity blocks are stored in the data buffer 230 (STEP 1004). ◆ STEP1
Rotate unit 4 of RAID management information 400 at 005
04, it is determined whether Ea is a rotate. ◆ If there is rotation, the process proceeds to STEP 1006, and the physical drive group is obtained from the changed physical drive ADR information 504 of the change management table 500 in order to find the physical drive group to store the changed logical data file.
See DR. Then, ADR information, redundancy degree 50
2. By inputting the division number 401 and the distribution unit 402 of the RAID management information 400 to the address conversion mechanism 280, the storage ADR after the redundancy is added is calculated. ◆ STEP100
At 7, go to address ADR obtained at STEP 1006 4
WR the data block / 2 parity block to the physical drive. For example, in the case of the RAID configuration as shown in FIG. 8, the logical data 15 is stored in the physical ADR2 of the physical drive 350a when the redundancy is 1. When the redundancy becomes 2, the physical drive 360a is changed to the physical ADR2.

When there is no rotate in STEP 1005, the process proceeds to STEP 1008, and the changed physical drive AD
By comparing the R information 504 and the physical drive ADR 405 of the RAID management information 400, the additional physical drive is searched and the additional parity is written (WR) to the additional physical drive. When the WR in STEP1007 / 1008 is completed, the process proceeds to STEP1009, and the storage end pointer 503 of the redundancy change information 500 is updated.
◆ The processing from STEP 1004 to STEP 1009 is repeated until the processing is completed for all the logical data files Ea.

When it is determined in STEP 1010 that the processing is completed, the process proceeds to STEP 1011 to change the redundancy degree 403 of the RAID management information 400 to the redundancy degree 502 of the redundancy change information 500, and the physical drive ADR 405 of the RAID management information 400. The physical drive ADR information 504 is changed after the redundancy change information 500 is changed. ◆ STE
FLAG5 being executed in the redundancy change information 500 in P1012
01 is reset. In STEP 1013, the data identifier of the additional physical drive ADR in the physical drive information 600 is changed from spare to data, and the extension process is completed.

The feature of this embodiment is that the RAID management information 40
0, change management table 500 and physical drive information 60
A parity generation circuit 260 having control information of 0 and capable of generating a plurality of redundant parity blocks is provided.
By performing the processing from 1004 to STEP 1009, it is possible to increase the number of redundant logical data files while the disk array apparatus and the host are online.

Next, a method of accessing the logical data file from the host when the changing process shown in FIG. 9 or 10 is being executed will be described with reference to the flowchart of FIG. ◆ When there is an access request from the CPU 100 (FIG. 1), the MPU 240 of the disk drive controller 200
(FIG. 2) is a running FLAG of the change management table 500 of the logical data file requested in STEP 1101.
See 501. ◆ Proceed to STEP 1102 and determine whether or not the change is in progress. If it is being changed, STEP110
Proceed to 3, and redundancy change information 5 of the target logical data file
The storage end pointer 503 of 00 is referred to.

Next, in STEP 1104, it is determined whether the logical data to be accessed has already been changed in redundancy. If it has been changed, the process proceeds to STEP 1105, the redundancy number 502 of the redundancy change information 500, the physical drive A after the change.
Number of divisions of DR information 504 and RAID management information 400 40
1, the distribution unit 402, and the rotate unit 404. ◆ If it is determined in STEP 1102 that the change is not in progress, or if it is determined in STEP 1104 that it has not been completed, the process proceeds to STEP 1106. And RAID
Reference is made to the division number 401, distribution unit 402, and rotation unit 404 physical drive ADR 405 of the management information 400.

The process proceeds from STEP 1105 or STEP 1106 to STEP 1107, and it is determined whether the request from the CPU 100 is a read request. ◆ When it is determined that the reading process is being performed, the process proceeds to STEP 1108. And
The information of STEP 1105 is input to the address conversion mechanism 280, and the physical drive storage AD of the logical data to be read is input.
Calculate R. ◆ Next, proceed to STEP 1110, STE
Data block from physical drive ADR of P1108
The parity block is read (RD) and stored in the data buffer 230. ◆ In STEP 1112, logical data is transferred to the host by removing the parity block from the data buffer.

When the write request is determined in STEP 1107, the information in STEP 1106 is input to the address conversion mechanism 280 to calculate the physical drive storage ADR of the logical data to be written (STEP 1109).
◆ Next, in STEP 1111, CP the write target data
The U100 receives the data in the data buffer 230, the parity generation circuit 260 generates the parity, and stores the parity in the data buffer 230. ◆ Next, in STEP 1113, the data block / parity block is written (WR) from the data buffer to the DRV.

As described above, according to this embodiment, the RAID
The management information 400, the change management table 500, the physical drive information 600, the parity generation circuit 260 capable of generating a plurality of parity blocks, STEP 1101-1 of FIG.
By means of 113, I / O from the CPU 100 can be accepted while the process of changing the redundancy of the logical data file is being executed, and the change process while the disk array device and the host are online is possible.

Further, according to the present embodiment, since the data block is once stored in the data buffer in the redundancy changing process and is subjected to ADR conversion and stored in the physical drive, it is read again from the physical drive after the storage is completed. By comparing the data on the data buffer, the reliability of the changed data can be improved.

Further, according to this embodiment, it is possible to automatically continue the changing process at the time of restarting after the power is shut off (P / S OFF) by the following method. ◆ In other words, the FLAG5 during execution of the change management table 500 at the time of restarting
It is determined whether 01 is set. If you are changing
By continuing the process from STEP 903 in FIG. 9, the change process can be automatically continued. ◆ This method uses the RAID management information 400 change management table 500,
Since the physical drive information 600 is set in the non-volatile memory, the power is cut off during the change (P / S OFF)
Even if it is done, it is executed because the information remains at the time of restarting. In this way, by always setting the interruption information of the change process in the non-volatile memory, the change process can be performed without the help of maintenance personnel after the power is turned on (P / S ON).

As a precaution, the methods used in the present invention will be listed and listed together with their characteristic points.

1) A disk drive device having a plurality of logical drive groups each of which includes a data storage drive and a plurality of physical drives composed of redundant data storage drives in one group; It has a data buffer for distributing or collecting the information and a non-volatile memory in part to intervene with the host device and control the transfer of information between them, and it can be used from the data that is further divided. In a disk array device including a disk drive control device having an ECC generation function for creating redundant data of variables, when an operator (user) requests reliability improvement for an arbitrary data file while online, the data From the data storage drive in which the logical drive group corresponding to the file is configured, Just read it into the data buffer and E
The number of redundant data created using the CC generation function is changed, the redundant data group created in the data buffer is stored in the redundant data storage drive from the data buffer, and the logical drive group existing in the nonvolatile memory is also stored. The number of redundant data of the logical drive group is increased by changing the redundant number or the management information for managing the storage target physical drive, and the reliability is improved during the operation (on) of the device. How to change the number of redundant data.

2) In the disk array device of the above 1), when a user requests reliability reduction for an arbitrary data file while online, the redundant number of logical drive groups existing in the non-volatile memory or the storage target physical drive. A method for changing the number of redundant data, characterized in that the number of redundant data in the logical drive group is reduced by changing the management information that manages the data, and the reliability is reduced while it is on.

3) In the disk array device of 1) above, when the number of redundant data in an arbitrary logical drive group is reduced as in 1) above, a physical drive for storing redundant data in the logical drive group is previously stored. When a drive failure occurs, refer to the above management information when assigning it as a spare drive according to the user's designation, by changing the spare drive management information that exists in the non-volatile memory. Recognizes the physical drive that is the spare drive, restores the data of the failed drive from the redundant data group that includes the physical drive, and stores it in the allocated spare drive, thus using the reduced physical drive as the spare drive. A redundant drive change management method characterized by the above.

4) In the disk array device of the above 1), when writing logical data from the upper level, the redundant number of the redundant data is referred by referring to the redundant number of the logical drive group in the nonvolatile memory or the management information of the physical drive to be stored. According to the above, redundant data is changed by generating redundant data using the ECC generation function and storing it in the target drive, so that the reliability of the data written from the host can be automatically changed after the redundant number is changed. Management method.

5) In the disk array device of the above 1), when reading the logical data from the higher order, the redundant number of the redundant data is referred by referring to the redundant number of the logical drive group in the nonvolatile memory or the management information of the physical drive to be stored. Data and redundant data are read into the data buffer according to the storage drive and the storage drive. When a failure occurs during reading, EC is read from the redundant data group read together with the data.
A redundant drive change management method characterized in that the reliability of data read from the host can be automatically changed after the redundancy number is changed by restoring the data using the C generation function and transferring the data to the host.

6) RAI for rotating and storing redundant data in a physical drive by the changing method in 1) above
In the D configuration, based on the logical data stored in the data buffer, redundant data for the redundant number newly changed by the ECC generation function is created, and the logical data and the redundant data group due to the change in the redundant data number A redundant drive change management method characterized in that a storage position is calculated using an address conversion mechanism, and a logical data group and a redundant data group are stored at physical drive addresses of respective storage target positions.

7) In the disk array device of 1) above,
R to store redundant data by rotating it to a physical drive
A redundant drive change management method characterized in that, in the AID configuration, the redundant data number reducing method in 2) above is also realized by the changing method in 5) above.

8) In the change processing of 1) above, when the number of redundant data is changed and generated by the ECC generation function and the redundant data is stored in the target physical drive,
The address of which storage has been completed is remembered in the non-volatile memory, and when there is a write request from the host to the logical data file being changed, the address is referred to,
If it is a write request to the area before the change, it will generate the redundant number of redundant data before the change and write it to the target drive.If it is a write request to the area after the change, it will write the redundant number of redundant data after the change. A redundant drive change management method characterized by receiving access from the host being changed and preventing performance degradation by generating and writing to the target drive.

9) In the change processing of the above 2), when a read request is issued from the host to the logical data file being changed, the address stored in the non-volatile memory is referred to read the area before the change. If it is a request, read the physical data making up the logical data and the redundant number of redundant data before the change from the target drive, and if it is a read request to the area after the change, the redundant number after the change A redundant drive change management method characterized by receiving access from the host being changed by reading data from the target drive to prevent performance degradation.

10) In the external storage device according to 1) above, in the changing method according to 8) or 9), when the redundant number is changed, the redundant data is stored in the changed redundant data storage drive. By remembering the completed pointer in the nonvolatile memory, the P / S of the disk array device
A redundant drive change management method characterized by automatically continuing the change process when restarting after being turned off.

11) In the disk array device of the above 1), before starting the change processing described in 7) or 8), the data in the logical data group to be changed is once read into the data buffer, and the change processing is performed. At the end of the process, the logical data is read from the physical drive that has re-stored the logical data to be changed, and the data before and after the change in the data buffer is compared to save the data for the online redundant data change process. Redundant drive change management method characterized by increasing reliability.

12) In the disk array device according to 1) above, the redundant drive to be stored when the number of redundant data according to 1) above is increased is assigned as a spare drive or is added online. Redundant drive change management method.

[0073]

According to the disk array device of the present invention, the redundancy of data can be changed during the operation of the disk array device, and the data file can be stored without stopping the disk array device (subsystem). Since the configuration can be changed, the reliability or performance of the entire device can be improved.

Further, even if the changing process takes a long time, the power can be cut off (P / S OFF) while the process is being executed. That is, at the next power-on (P / S ON), the change processing can be automatically restarted from the time of interruption without intervention of maintenance personnel, which leads to improvement in maintenance performance and usability.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a computer system including a disk drive device according to an embodiment of the present invention.

FIG. 2 is a conceptual diagram showing in more detail a part of a disk drive device according to an embodiment of the present invention.

FIG. 3 is a diagram showing a drive configuration of a disk array device that constitutes a disk drive device according to an embodiment of the present invention.

FIG. 4 is a conceptual diagram showing RAID management information used in a disk array device which is an embodiment of the present invention.

FIG. 5 is a conceptual diagram showing a change management table used in the disk array device which is an embodiment of the present invention.

FIG. 6 is a conceptual diagram showing physical drive information used in a disk array device which is an embodiment of the present invention.

FIG. 7 is a conceptual diagram showing a data format of logical data used in the disk array device which is an embodiment of the present invention.

FIG. 8 is a view showing rotation (the number of divisions =) in the processing of changing the redundancy of the disk array device according to the embodiment of the present invention.
4 is a conceptual diagram showing an example of a data storage format for changing the redundancy from 1 to 2 in the case of 4 / distribution unit = 2 / rotate unit = 2).

FIG. 9 is a flow chart showing a method of controlling a disk array device which is an embodiment of the present invention.

FIG. 10 is a flow chart showing a control method of the disk array device which is an embodiment of the present invention.

FIG. 11 is a flowchart showing a control method of the disk array device which is an embodiment of the present invention.

[Explanation of symbols]

100; Central processing unit (CPU), 200; Disk drive control device, 210; Channel control device, 2
20; drive control device, 230; data buffer, 2
40; Microprocessor unit (MPU), 25
0; RAM, 260; parity generation circuit, 261; 1 parity generation, 262; 2 parity generation, 270; non-volatile memory, 280;
Address translation mechanism, 300; disk drive device,
310-360; data transfer control circuit, 310a-36
0a; physical drive (logical data file Ea), 31
0b to 360b; physical drive (logical data file E
b), 310c to 360c; physical drive (logical data file Ec), 310d to 360d; physical drive (logical data file Ed), 400; RAID management information, 401; division number, 402; distribution unit,
403; redundancy degree, 404: rotation unit,
405; Physical drive ADR, 500; Change management table, 501; Executing FLAG, 502;
Redundancy 502, 503; Storage end pointer 5
03, 504; additional physical drive ADR 504, 60
0: Physical drive information, 601: Logical data file number, 602: Data identifier, 800: ECC group (parite generation unit), 801-804; Logical data block, 805-806; Parity block, 901
~ 913; redundancy reduction processing flow, 1001 to 101
3; redundancy increase flow, 1100 to 1113; host access processing flow.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification number Reference number within the agency FI Technical indication location G11B 20/18 574 E 8940-5D (72) Inventor Takao Sato 2880, Kozu, Odawara-shi, Kanagawa Stock Association Storage Systems Division, Hitachi, Ltd.

Claims (5)

[Claims] 1. A disk array device capable of changing data redundancy while being online connected to a host device or without shutting off its own power source. 2. The disk array device according to claim 1,
Furthermore, the disk array device that executes the input / output processing for the logical data file of the host device while changing the redundancy. 3. The disk array device according to claim 1,
Furthermore, the disk array device which continues the change of the redundancy when the power is turned on after the power of itself is cut off during the change of the redundancy. 4. The disk array device according to claim 1,
Furthermore, a disk array device that indicates that the redundancy is being changed. 5. A parity generation circuit for generating parity data of one or a plurality of redundancy levels, write request data from a host, a data block or parity block from a physical drive, or 1 generated from the parity generation circuit. A data buffer that stores one or more parity blocks, at least the RAID of the specified logical data file
From the RAID management information that manages the configuration, the redundancy change information requested by the operator and the information of the data identifier indicating the usage of the physical drive, and the storage address of the logical data and the RAID management information, A disk array device having an address conversion mechanism for calculating the address of a storage target physical drive and the address of physical drive data.