JPH0744331A - Disk array device and its control method - Google Patents

Abstract

PURPOSE:To simultaneously write data into plural disk array devices at a high speed and also to improve the reliability of these array devices. CONSTITUTION:In a normal operation state, a pair disk 31 is used as a copy of a reundant disk 21. Then both disks 21 and 31 are used for different data disks to write data when the data are read out and written into the data disks 11-14 and the disk 21. Thus the data can be simultaneously written into plural disks. Then the data written in both disks 21 and 31 are copied to each other after the data writing processes are completed. In such a way, the disk 31 is used as a copy of the disk 21 so that the load applied to the disk 21 is scattered in a read modified write mode.

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 having a spare disk and a control method thereof, and more particularly, by using the spare disk as a copy disk of a redundant disk, realization of simultaneous multiple write processing and its speedup. , And to a disk array device and a control method therefor capable of improving the reliability of the disk array device.

[0002]

2. Description of the Related Art In recent information-oriented society, the amount of information to be processed is increasing, and there is a strong demand for speeding up of information processing devices. Although the processing speed of the computer main body has been increased by technological development, the operating speed of the external storage device (particularly the hard disk device) cannot keep up with the processing speed of the computer main body, which has been a bottleneck in improving the throughput of the entire processing. Although a single hard disk device itself has been improved in order to improve the performance of the hard disk device, a method of improving the performance as a whole by combining a plurality of hard disk devices has been considered. One of the methods is a disk array device which is the premise of the present invention.

The disk array device was proposed by DA Patterson et al.
edundant Arrays of Inexpensive Disks), which is a system that uses multiple hard disk drives to realize a hard disk drive with higher reliability and higher performance than a single hard disk drive. It has been used, but recently, it is a technology that has been drawing attention as an external storage device for personal computers and workstations. This device has a plurality of built-in hard disk devices, and is accessed from a host computer as a single hard disk device. A disk array device consisting of multiple hard disk devices and a single hard disk device are similar to the relationship between a multiprocessor system and a single processor system. High performance can be achieved by using multiple devices as in the case of a multiprocessor system. Software support is essential to withdraw.

Regarding the disk array device, for example,
It is explained in detail in the following papers. 1: D.Patterson, G.Gibson, and RHKatz "A Case For Red
undant Arrays of Inexpensive Disks (RAID) "[Proceed
ings of the 1988 ACM Conference on Management of D
ata (SIGMOD), Chicago, IL., June 1988, pp.109-116] 2: "Nikkei Electronics 1993-426 (No.579)"
(Nikkei BP, issued April 26, 1993) pp.77-103 The disk array device divides the input data,
The divided data and the error correction code are distributed and stored in a plurality of hard disk devices. In the above paper, the disk array device has five levels (RAID1 to RAID
It is divided into 5) and its performance and reliability are discussed.

First, the five levels will be briefly described. Level 1 (RAID 1) is a duplication of a hard disk device (mirrored disk). Level 2 (RA
The ID2) and the level 3 (RAID3) are for dividing (striping) the input data in bit units, and interleaving the data into a plurality of hard disk devices for storage. Level 2 (RAID2) uses a Hamming code as an error correction code, and when there are four data storage hard disk devices, three error correction code hard disk devices (redundant disks) are required. On the other hand, in level 3 (RAID 3), parity is used as an error correction code, and even when there are four data storage hard disk devices, only one error correction code hard disk device (redundant disk) is required.

Level 4 (RAID 4) and Level 5
In (RAID 5), input data is divided (striped) into sector units, interleaved and stored in a plurality of hard disk devices, and parity is used as an error correction code. Level 4 (RAID4) uses a single hard disk drive for parity (redundant disk)
However, in level 5 (RAID 5), the parity is distributed and stored in a plurality of hard disk devices.

Level 4 (RAID 4) and Level 5
In the (RAID 5) disk array device, the independence of each data disk can be increased by increasing the stripe size (division size). Therefore, it becomes possible to simultaneously access a plurality of data. Specifically, in level 4 (RAID4), simultaneous multiple read processing is possible, and in level 5 (RAID5), both simultaneous multiple read processing and simultaneous multiple write processing are possible.

However, in order to perform the write processing in the disk array device, the read modify write processing must be performed. For that purpose, it is necessary to access the disk storing the redundant data. Therefore, if the redundant data of the two data for which the simultaneous write processing is attempted is stored in the same disk, they cannot be processed simultaneously. Therefore, a level 4 (RAID4) disk array device having one redundant disk cannot perform simultaneous multiple write processing. For the same reason, level 5 (RAID
Even in the disk array device of 5), it is not always possible to perform simultaneous multiple write processing. As a technique for improving the reliability of a disk array device, a method using a spare disk has been widely known. However, in the conventional method, since the spare disk is provided as a spare disk for a disk failure, it is used when a failure occurs, but is not used at all in a normal usage state where no failure has occurred and is useless. Disk, the usage efficiency is not so good when viewed as a disk array device as a whole.

A conventional general disk array device is shown in FIG. FIG. 5 shows a configuration example of a disk array device (disk array device including four data disks, one redundant disk, and one spare disk) including a conventional spare disk and a redundant disk with a buffer memory. In FIG. 5, 1 is a host computer, 2
Is a disk array controller, 11 is the first data disk, 12 is the second data disk, 13 is the third data disk, 14 is the fourth data disk, 21
Is a redundant disk, 31 is a spare disk, and 87 is a write buffer for the redundant disk.

[0010]

As described above, in order to perform the write processing in the disk array device, it is necessary to access both the data disk and the redundant disk. Therefore, in order to perform the simultaneous multiple write processing in the disk array device, the data disks to be accessed by the multiple write commands executed simultaneously and the redundant disks must all be different. Therefore, in the level 4 (RAID4) disk array device, since only one disk stores redundant data, simultaneous multiple write processing cannot be performed. Furthermore, in a level 5 (RAID 5) disk array device, simultaneous multiple write processing is possible only when the data disk to be accessed by a plurality of write commands to be executed at the same time and the redundant disk are all different. Or, since there is a possibility of contention for redundant disks, simultaneous multiple write processing cannot always be performed.

The prior art for improving reliability using a spare disk is used only when a failure occurs because the spare disk is configured to work as a spare only when a failure occurs as described above. However, the hardware (spare disk) is not used effectively during normal times (when there is no failure), which takes up most of the time, and there is the problem that the disk device is not used for a long period of time. If this happens, there is a problem that rust may be generated in the internal mechanical portion of the disk device, making it impossible to use the disk device when trying to use it. The purpose of the present invention is to
By improving such problems, the effective use of spare disks during normal times (when there is no failure) and the number of disks that store redundant data are increased to reduce disk contention during simultaneous multiple write processing and to enable simultaneous multiple access. It is an object of the present invention to provide a disk array device that can be realized and speeded up and a control method thereof.

[0012]

In order to achieve the above object, the apparatus of the present invention comprises a plurality of data disks storing simultaneously accessible data and an error correction code for the data stored in the data disks. Read-modify-write to the data disk and redundant disk to be written at the time of data write processing, and perform read processing only from the disk that stores the data to be read at the time of data read processing. In the disk array device, a spare disk is further provided, and the spare disk is used as a spare disk when a data disk or a redundant disk fails, and used as a redundant disk copy disk during normal operation.

The control method of the disk array device of the present invention comprises a plurality of data disks storing simultaneously accessible data and a redundant disk storing the error correction code of the data stored in the data disks. In the control method of the disk array device, the read processing is performed only from the disk in which the data to be read is stored during the data read processing, and the read modify write is performed to the data disk and the redundant disk to be written during the write processing to the disk. , A spare disk is provided, and the spare disk is used as a spare disk when a data disk or a redundant disk fails, and is used as a copy disk of the redundant disk during normal operation. Read and write of the redundant disk and the spare disk together with the write processing of the separate data disks to perform multiple simultaneous write processing, and after the write processing is completed, the data written to the redundant disk and the spare disk are copied to each other. ing.

Further, when simultaneously writing a plurality of data disks and redundant disks, write buffers dedicated to the redundant disks and spare disks are provided, and the write buffers store the redundant data to be written. The redundant data is written to the corresponding disk, and after the processing is completed, each redundant data stored in the write buffer is copied to the other disk. In addition, a spare disk and redundant disk compare check is performed between disk accesses,
If the check results do not match, the data disk is read, the redundant data is generated, and it is determined which of the spare disk and the redundant disk has the correct redundant data,
The redundant data of the wrong one is updated to the correct data. When a data disk fails, the data of the failed disk is regenerated from the remaining data disks and redundant or spare disks, and the generated data is stored in either the redundant disk or the spare disk. I am trying to recover the failed disk. If a spare disk or redundant disk fails, disconnect the failed disk,
When a new spare disk is connected instead of the failed disk, the contents of the redundant disk are copied to the spare disk.

[0015]

According to the present invention, a spare disk is provided in the disk array device, and the spare disk is used as a spare disk when a data disk or a redundant disk fails, and as a redundant disk copy disk during normal operation. , Has the following effects.
In a level 4 (RAID4) disk array device, since the same redundant data is stored in different disks during normal operation, a plurality of write processes can be performed simultaneously unless the data disks to be accessed conflict. In a level 5 (RAID 5) disk array device, since the same redundant data is stored in different disks during normal operation, multiple write processes can be performed at the same time unless the data disk to be accessed and the redundant disk conflict. it can. Further, since redundant data is multiplexed and the spare disk is used as a redundant disk even in normal times, the internal mechanical portion of the spare disk is less likely to rust and the reliability of the disk array device is improved.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a disk array device (4 data disks, redundant disk 1 according to the first embodiment of the present invention.
Level 4 (RAID
It is a block diagram of (4) disk array device). In FIG. 1, 1 is a host computer, 2 is a disk array controller, 11 is a first data disk, 12 is a second data disk, 13 is a third data disk, 1
4 is the fourth data disk, 21 is a redundant disk, 3
Reference numeral 1 is a spare disk. Reference numeral 87 is a write buffer for a redundant disk, and 88 is a write buffer for a spare disk.

In the first embodiment, during normal use, the spare disk 31 stores the same data as the redundant disk 21. FIG. 2 shows an example of an array of data and redundant data of each disk in a level 4 (RAID4) disk array device. In the case of the read processing, the data disk storing the data to be read is accessed and read, as in a normal disk array device.
Next, the writing process will be described. It is assumed that the host computer 1 has issued an instruction to perform a write process on one block of data D12 and one block of data D23. In order to perform the writing process, it is necessary to access the data D12 and the parity data P2 in the same row as the data D12, and the data D23 and the parity data P3 in the same row as the data D23.

In the conventional disk array device, since the parity data P2 and the parity data P3 are on the same disk (parity disk 21), two data (D12
And D23) could not be written at the same time. However, in the disk array device of the present invention, since the spare disk 31 stores exactly the same data as the redundant disk 21, use P2 in the spare disk 31 instead of P2 in the parity disk 21. As a result, D12, P2, D23, and P3 are all stored in different disks (data disk 11, spare disk 31, data disk 12, redundant disk 21) and can be accessed at the same time.

The write processing (read modify write) in this embodiment is performed in the following procedure. (As an example, consider the write processing of the data D12 and the data D23.) First, the data disk 11 and the spare disk 31 are accessed, and the read modify write processing of the data D12 and the parity data P2 of the same row as the data D12 is performed. At the same time, the data disk 12 and the redundant (parity) disk 21 are accessed to perform the read-modify-write processing of the data D23 and the parity data P3 of the same row as the data 23. Next, writing D12 (new data) to the data disk 11, writing P2 (new parity) to the spare buffer buffer 88, writing D23 (new data) to the data disk 12, and P3 (new parity). Writing to the redundant (parity) disk buffer 87. After that, the data P2 in the buffer memory 88 is written to the spare disk 31, and the data P3 in the buffer memory 87 is written to the redundant (parity) disk 21. Further, thereafter, the data P2 in the buffer memory 88 is transferred to the redundant (parity) disk 2
1 and the data P3 in the buffer memory 87 is written in the spare disk 31.

Further, a compare check of the spare disk and the redundant disk is executed between the disk accesses, and if the check results do not match, the data disk is read out to generate the redundant data, and the spare disk and the redundant disk are checked. It is determined which has the correct redundant data and the incorrect redundant data is updated to the correct data.

Next, in this embodiment, level 4 (RA
A processing procedure when a failure occurs in each of the data disk, the redundant disk, and the spare disk that configure the disk array device of ID4) will be described. Level 4
In the (RAID4) disk array device, when a data disk fails, the data of the failed disk is regenerated by the data of the remaining data disk and the redundant disk 21, and the generated data is stored in the spare disk 31. Store and recover the contents of the failed disk. After that, when a new spare disk is connected instead of the failed disk, the data of the redundant disk is copied to the spare disk.

In the level 4 (RAID4) disk array device, when a failure occurs in the redundant disk 21, the failure disk (redundant disk 21) is disconnected,
A new disk is connected instead of the failed disk, and the contents of the spare disk 31 are copied to the disk. Further, in the level 4 (RAID4) disk array device, when a failure occurs in the spare disk 31,
The failed disk (spare disk) is separated, a new disk is connected instead of the failed disk, and the contents of the redundant disk 21 are copied to the disk.

Next, a second embodiment of the present invention in a level 5 (RAID 5) disk array device will be described.
FIG. 3 is a configuration diagram of a disk array device (a level 5 (RAID5) disk array device including five data disks and one spare disk) according to the second embodiment. In FIG. 3, 1 is a host computer, 2 is a disk array control device, 11 is the first data disk,
Reference numeral 12 is a second data disk, 13 is a third data disk, 14 is a fourth data disk, 15 is a fifth data disk, and 31 is a spare disk. 8
1 is a buffer for the first data disk 11, 82 is a buffer for the second data disk 12, 83 is a buffer for the third data disk 13, and 84 is a buffer for the fourth data disk 14. Buffer, 85 is a buffer for the fifth data disk 15, 88 is a spare disk 3
This is a buffer for 1.

In the second embodiment, during normal use, the spare disk 31 has the same redundant data as the redundant data (P1 to Pn) distributed and stored in the data disks (11 to 15). It is stored at the address. FIG. 4 shows an example of an array of data and redundant data (P1 to Pn) of each data disk (11 to 15) in the level 5 (RAID5) disk array device. In the case of read processing, the data disk storing the data to be read is accessed, as in a normal disk array device. Next, the writing process will be described. It is assumed that the host computer 1 has issued an instruction to perform a write process on one block of data D12 and one block of data D23. In order to perform the writing process, the data D12 and the parity data P2 in the same row as the data 12 are written.
Further, the data D23 and the parity data P3 in the same row as the data 23 must be accessed.

In the conventional disk array device, since the parity data P2 and the data D23 are on the same disk (data disk 12), two data (D12 and D2) are stored.
It was not possible to write 3) at the same time. However,
In the disk array device of the present invention, the spare disk 31
Since the same redundant data as the redundant data (P1 to Pn) distributed and arranged in each data disk (11 to 15) is stored at the same address, the spare data is stored in place of P2 in the data disk 12. By using P2 in the disk, D12, P2, D23, and P3 are all different disks (data disk 11, spare disk 31, data disk 12, data disk 13).
It will be stored in and can be accessed at the same time.

The write processing (read modify write) in this embodiment is performed in the following procedure. (As an example, consider the write processing of the data D12 and the data D23.) First, the data disk 11 and the spare disk 31 are accessed, and the read modify write processing of the data D12 and the parity data P2 of the same row as the data D12 is performed. At the same time, the data disk 12 and the data disk 13 are accessed to access the data D23 and the data 2
The read-modify-write process of the parity data P3 in the same row as that of No. 3 is performed. Next, writing D12 (new data) to the data disk 11, writing P2 (new parity) to the spare buffer buffer 88, writing D23 (new data) to the data disk 12, and P3 (new parity). Writing to the buffer 83 for the data disk 13. After that, the data P in the buffer memory 88
2 is written to the spare disk 31 and the buffer memory 8
The data P3 of No. 3 is written on the data disk 13. Further, after that, the data P2 in the buffer memory 88 is written to the data disk 13, and the data P2 in the buffer memory 83 is written.
3 is written in the spare disk 31.

Further, a compare check of the redundant data in the spare disk and the redundant data in the data disk is executed between the disk accesses, and if the check results do not match, the data disk is read out to generate the redundant data. Then, it is determined which of the spare disk and the data disk has the correct redundant data, and the incorrect redundant data is updated to the correct data.

Next, a processing procedure when a failure occurs in each of the data disk and the spare disk which constitute the level 5 (RAID5) disk array device will be described. In a level 5 (RAID 5) disk array device, when a data disk fails, the data of the failed disk is regenerated from the remaining data disks, and the generated data is stored in the spare disk 31. Recover the failed disk. After that, if a new disk is connected instead of the failed disk,
Copy redundant data to the disk.

In the level 5 (RAID 5) disk array device, when a failure occurs in the spare disk 31, the failure disk (spare disk 31) is disconnected, and if a new disk is connected instead of the failure disk, Copy redundant data to the disk.

[0030]

According to the present invention, level 4 (RAID
It is possible to realize the simultaneous multiple write processing in the disk array device of 4) and to speed up the simultaneous multiple write processing in level 5 (RAID 5). Further, since redundant data is multiplexed and the spare disk is used even during normal operation, the internal mechanical section is less likely to rust, and the reliability of the disk array device can be improved.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a disk array device according to a first embodiment of the present invention.

FIG. 2 shows a distribution state of data and redundant data in the disk array device according to the first embodiment of the present invention.

FIG. 3 is a configuration diagram of a disk array device according to a second embodiment of the present invention.

FIG. 4 is a diagram showing a distribution status of data and redundant data in a disk array device according to a second embodiment of the present invention.

FIG. 5 is a configuration diagram of a disk array device including a conventional spare disk and a buffer memory for a redundant disk.

[Explanation of symbols]

 1 Host Computer 2 Disk Array Controller 11 1st Data Disk 12 2nd Data Disk 13 3rd Data Disk 14 4th Data Disk 15 5th Data Disk 21 Redundant Disk 31 Spare Disk 81 Buffer memory for the first data disk 82 Buffer memory for the second data disk 83 Buffer memory for the third data disk 84 Buffer memory for the fourth data disk 85 Buffer memory for the fifth data disk 87 Buffer memory for redundant disk 88 Buffer memory for spare disk

Claims (6)

[Claims] 1. A plurality of data disks for storing simultaneously accessible data and a redundant disk for storing an error correction code of the data stored in the data disks, and stores data to be read during a data reading process. In the disk array device that performs read processing only from the disk that is being written and performs read-modify-write to the data disk to be written and the redundant disk at the time of write processing to the disk, a spare disk is further provided and A disk array device characterized by being used as a spare disk when a disk or redundant disk fails and used as a disk for copying the redundant disk during normal operation. 2. A plurality of data disks for storing simultaneously accessible data and a redundant disk for storing an error correction code of the data stored in the data disks, wherein data to be read is stored during a data reading process. In the control method of the disk array device in which the read processing is performed only from the disk that is written, and the write processing to the disk is performed, the spare disk is provided in the control method of the disk array device that performs read-modify-write to the data disk and the redundant disk to be written
Used as a spare disk when the data disk or the redundant disk fails, and used as a copy disk for the redundant disk during normal operation.Reading and writing to the data disk and the redundant disk is performed by the redundant disk and the spare disk. A disk array device, characterized in that the disks are used together with the write processing of different data disks to perform a plurality of simultaneous write processing, and after the write processing is completed, the data written to each of the redundant disk and the spare disk is copied to each other. Control method. 3. The method of controlling a disk array device according to claim 2, wherein when the data disk and the redundant disk are simultaneously written to a plurality of times, write buffers dedicated to the redundant disk and the spare disk are respectively provided and the write operation is performed. The buffer stores redundant data for write processing, writes the redundant data to the corresponding disk, and copies the redundant data stored in the write buffer to the other disk after the processing is completed. Disk array device 4. The method of controlling a disk array device according to claim 2, wherein a compare check between the spare disk and the redundant disk is executed between disk accesses, and if the check results do not match, the data disk is deleted. It is characterized by reading out, generating redundant data based on that data, determining whether the spare disk or the redundant disk has the correct redundant data, and updating the incorrect redundant data to the correct data. Disk array device control method. 5. The method of controlling a disk array device according to claim 2, wherein when a failure occurs in the data disk, the failure disk data is deleted based on the contents of the remaining data disk and redundant disk or spare disk. A method of controlling a disk array device, wherein the data is regenerated, the generated data is stored in either a redundant disk or a spare disk, and the failed disk is recovered. 6. The method of controlling a disk array device according to claim 2, wherein when a failure occurs in the spare disk or the redundant disk, the failed disk is separated and a new spare disk is connected instead of the failed disk. After that, the contents of the redundant disk are copied to the spare disk, and the method of controlling the disk array device.