JPH10240453A - Disk array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a disk array device with which reliability is more improved by duplexing a parity. SOLUTION: A disk device 2 for parity similar to an RAID 3 is provided, sectors (n) for recording parities PP1-PP4 of data recorded in sectors from 0 to n-1 for data are also provided in disk devices from 1-1 to 1-4 for data, and the parity is duplexed. When extremely high reliability is requested, a 2nd reading part 44 parallelly accesses the same sectors of disk devices from 1-1 to 1-4 and 2, performs a parity check based on data read out of disk devices from 1-1 to 1-4 and the parity read out of the disk device 2, reads out the parities PP1-PP4 in disk and data of sectors as generation sources for each of respective disk devices from 1-1 to 1-4 and performs a parity check based on them.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array apparatus for performing data input / output processing by accessing a plurality of disk apparatuses in parallel, and more particularly to a disk array apparatus having a dual parity to improve reliability. .

[0002]

2. Description of the Related Art As a disk array device, RAID is used.
Although various configurations such as 1 to RAID 5 have been proposed, a system mainly executing an application for continuous access to a large amount of data is capable of high-speed transfer of large-capacity data and using one redundant disk. Finished RAID3 is suitable.

FIG. 6 shows a configuration example of a conventional RAID 3 disk array device. In FIG.
1-4 is a disk device for storing data, 102 is a disk device for storing parity, which constitutes a disk array 100, a disk array controller 104 is connected to the disk array 100, and a disk array controller 104 A computer 103 as a higher-level device is connected to the device 104. When writing data a, b, c, d as shown in FIG. 2 to this disk array device,
The disk array controller 104, the data a ₀ ~a ₃ sector _{units, b 0 ~b 3, c 0} ~c 3, d 0
Ｄd ₃ , the parity P0 is calculated from the data a _{0 to} a ₃ , the parity P1 is calculated from the data b _{0 to} b ₃ ,
Parity P2 is calculated from the data c ₀ to c _3, data d
Parity P3 is calculated from ₀ to d _3, the disk device 10
1-1 to 101-4 and 102 are accessed simultaneously in parallel, and as shown in FIG.
Write to. In addition, data reading is performed, for example, by taking data a as an example, using disk devices 101-1 to 101-4.
And 102 are accessed in parallel and data a _{0 to} a
₃ and the parity P0 are read, a parity check is performed, and if there is no error, the data a is assembled and output to the computer 103. The identification of a failed disk in the case of an error is performed by the control device according to a unique detection method of each disk, and the error is corrected if possible.

[0004] As described above, in RAID3, since all the disk devices operate in parallel for both data writing and data reading, high-speed access to large-capacity data is possible.

[0005]

SUMMARY OF THE INVENTION By the way, RAID3
Is suitable for continuous transfer of large-capacity data as described above, and has the advantage that only one redundant disk is required regardless of the number of parallelized disks. Since the redundant information is stored in one parity, it is still insufficient for storing data requiring extremely high reliability.

SUMMARY OF THE INVENTION An object of the present invention is to provide a disk array device in which the reliability is further improved by duplicating the parity.

[0007]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a disk array device which accesses a plurality of disk devices in parallel to perform data input / output processing, and records data in sectors of its own disk. A plurality of first disk devices for recording parity of the obtained data in a specific sector of the own disk as parity in the disk, and a method for recording each sector in a sector at the same position of the plurality of first disk devices. A disk array configured with a second disk device used for storing parity of the
A disk array control device that performs input / output processing of data and parity with respect to the disk array.

The disk array control device divides the data into sector units and writes the data in parallel with the parity in the same sector of the first and second disk devices, and the first data to be changed by the writing. A writer for rewriting parity in the disk of the disk device;
And accessing the same sector of the second disk device in parallel and performing a parity check based on the data of the sector read from the plurality of first disk devices and the parity of the same sector read from the second disk device. 1
And a read unit that accesses the same sector of the first and second disk devices in parallel at the time of a data read request specifying the second mode, and reads the data of the sector read from the plurality of first disk devices and the second sector. A parity check is performed based on the parity of the same sector read from the disk device, and the parity in the disk and the data of the sector from which the parity is generated are read for each first disk device, and the parity check is performed based on the parity. 2 reading units.

[0009] Each of the first disk devices may be provided with one sector for storing the in-disk parity, and one sector for storing the in-disk parity for every several consecutive sectors. May be provided.

As described above, in the disk array device of the present invention, the parity of the data recorded in each sector is held in the disk device for storing data in addition to the disk device for storing parity. 2 using parity
Perform a heavy check.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, embodiments of the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of a disk array device according to the present invention. In the figure, 1-1 to 1-4 are first disk devices, 2 is a second disk device,
These constitute the disk array 1. Also, 4
Is a disk array control device connected to the disk array 1, and 3 is a computer as a host device.

The first disk devices 1-1 to 1-4 have a plurality of sectors from sector 0 to sector n. Sectors 0 to n-1 are used for recording data, and sector n is used for sectors 0 to n.
It is used for recording parity (in-disk parity) PP1 to PP4 of data recorded in n-1. On the other hand, the second disk device 2 has a plurality of sectors from sector 0 to sector n, and the sectors 0 to n correspond to the data recorded in the same sector of the first disk devices 1-1 to 1-4. , Pn of the parity P0, P1,..., Pn. The plurality of disk devices 1-1 to 1-4, 2 are RAID
3 and operate in parallel at the same time.

The disk array control device 4 is a device for controlling the input / output processing of data and parity to and from the disk array 1. As shown in the figure, a disk access unit 41 for controlling access to the disk array 1, a writing unit 42, A first reading unit 43 and a second reading unit 4
4 and an interface unit 45 for the computer 3. These are program controlled CPUs
And a memory or the like.

The interface unit 45 is provided with the computer 3
From the disk array device, and decodes the command,
At the time of writing, control is passed to the writing unit 42, and at the time of reading, the first reading unit 43 or the second reading unit 4 is read.
Pass control to 4. Which readout unit to pass control to depends on
It depends on the read mode from the computer 3. That is, the first read unit 43 is activated when a read request specifying the first mode is performed, and the second read unit 44 is activated when a read request specifies the second mode.

The write unit 42 divides the write data into sector units, and divides the write data together with its parity into each disk device 1.
A process for writing data in parallel to the same sectors -1 to 1-4 and 2 and rewriting parity in the disks of the first disk devices 1-1 to 1-4 to be changed by the writing are performed.

The first read unit 43 accesses the same sector of each of the disk devices 1-1 to 1-4 and 2 in parallel, and reads the sectors read from the first disk devices 1-1 to 1-4. A parity check based on the data and the parity of the same sector read from the second disk device 2 is performed, and if there is no error, the read data is transferred to the computer 3 through the interface unit 45.

The second read unit 44 accesses the same sector of each of the disk devices 1-1 to 1-4 and 2 in parallel, and reads the sectors read from the first disk devices 1-1 to 1-4. A parity check based on the data and the parity of the same sector read from the second disk device 2 is performed, and the parity in the disk and the source of the parity are generated for each of the first disk devices 1-1 to 1-4. The data of the sector is read, a parity check is performed based on the data, and if there is no error, the read data is transferred to the computer 3 through the interface unit 45.

When an error is detected in the first reading section 43 or the second reading section 44, the conventional RA is used.
If possible, the data is corrected as in the case of ID3.

Next, the operation of this embodiment will be described for each case. In the following description, old data and parity are distinguished by prefixing "old" characters, and new data and parity are prefixed by "new" characters. For example, when writing data a ₀ to sector 0 of the disk device 1-1, the original data is old a _0, trying to write data referred to as new a _0.

(1) Writing of Data For example, as shown in FIG. 2, when writing data a, b, c, and d to sectors 0, 1, 2, and 3, the writing unit 42 writes each data into a new a _{0 to} a ₃ , new b _{0 to} b
_3, the new c ₀ to c _3, divided into new d ₀ to d _3, the new a ₀ ~a
The new parity P0 is calculated from the _3, new b ₀ ~b the new parity P1 is calculated from the _3, new c ₀ ~c ₃ new parity from P2
The calculated, calculates the new parity P3 from the new d ₀ ~d _3,
Sectors 0, 1, and 1 of the disk devices 1-1 to 1-4 and 2
2 and 3 are sequentially accessed and written in parallel. This operation is basically the same as that of the conventional RAID3. However,
Since it is necessary to update the parity PP1 to PP4 in the old disk of the sector n of each disk device 1-1 to 1-4,
Specifically, data writing is performed, for example, according to the procedure shown in FIG.

First, the writing unit 42
Parity PP1 in the old disk of the sector n of -1 to 1-4
To PP4 are read in parallel and held internally (S
1). Next, the old a ₀ of the sector 0 to which the data a is written
The ~a ₃ parallel read from the disk device 1-1 to 1-4 (S2). Then, with the former a ₀ ~a ₃ thus read out,
Parities PP1 to PP4 stored in the old disk,
From the new a ₀ ~a ₃ Prefecture, new disk parity PP1~P be written to the sector n of the disk device 1-1 to 1-4
The value of P4 is calculated and held as a new parity in the old disk (S3). Then, write the new a ₀ ~a ₃ to write the sectors 0 new parity P0 which is the parity in parallel (S4). Then, returning to step S2, data b is similarly written, and thereafter, data c and data d are similarly written. When the writing of the data d is completed, the writing data is completed (S5), and the retained parity PP1 to PP4 in the old disk and the parity Pn are stored in the sector n of each disk device 1-1 to 1-4, 2. Write in parallel (S6).

(2) Reading in the first mode For example, when reading data a, the first reading unit 43
Reads the data a _{0 to} a ₃ and the parity P 1 from the sector 0 by simultaneously accessing the disk devices 1-1 to 1-4 and 2 in parallel, and performs a parity check. If there are no errors, output from the data a ₀ ~a ₃ to the computer 3 assembles the data a.

(3) Reading in the second mode For example, when reading data a, the second reading unit 44
Reads the data a _{0 to} a ₃ and the parity P 1 from the sector 0 by simultaneously accessing the disk devices 1-1 to 1-4 and 2 in parallel, and performs a parity check. If there is no error, the disk devices 1-1 to 1-4 are simultaneously accessed in parallel to read data of all the remaining sectors and perform a parity check. The parity check performed at this time is a check closed to each of the disk devices 1-1 to 1-4. In the disk device 1-1, the parity generated by the data from sector 0 to sector n-1 is the parity of sector n. If it matches the parity PP1, there is no error. The same applies to the other disk devices 1-2 to 1-4. If even there is no error in any of the disk device, and outputs to the computer 3 from the data a ₀ ~a ₃ read previously assembled data a. The parity P read from sector n of the disk devices 1-1 to 1-4 is
It may be checked whether the parity generated from P1 to PP4 matches the parity Pn recorded in the sector n of the disk device 2.

FIG. 4 is a block diagram of another embodiment of the disk array apparatus according to the present invention, and the same reference numerals as those in FIG. 1 denote the same parts. In the embodiment of FIG. 1, one sector for storing parity in the disk is provided as one sector n in the disk devices 1-1 to 1-4, but in this embodiment, a predetermined number ( In the figure, one sector for storing parity in the disk is provided for each physically continuous sector 4). The operation of this embodiment is as follows.

(1) Writing of Data For example, when writing data a and b among the data a, b, c and d shown in FIG.
In a new a ₀ ~a ₃ each data on a sector-by-sector basis, new b
_{0 to} b ₃ , a new parity P0 is calculated from the new a _{0 to} a ₃ , a new parity P1 is calculated from the new b _{0 to} b _3, and sectors 0 of the disk devices 1-1 to 1-4 and 2 are calculated. , 1 are sequentially accessed and written in parallel. This operation is basically the same as that of the conventional RAID3. However, data a, b
Since the sector 4 holds the parity in the disk,
It is necessary to update the old disk parities PP11 to PP41 of the sector 4 of each of the disk devices 1-1 to 1-4 and 2. For this reason, writing of data is specifically performed by a procedure as shown in FIG. 5, for example.

First, the writing section 42 writes the first data a
The sector of the parity in the disk to be changed is obtained by writing (S10). In this embodiment, since one in-disk parity sector is provided for every four consecutive sectors, it can be obtained from the number of the sector in which data a is written. In this case, it is sector 4. Next, the old parities PP11 to PP41 of the sectors 4 of the disk devices 1-1 to 1-4 are read out in parallel and held inside (S11). Then, the old a ₀ ~a ₃ disk drive sector 0 of the write destination of data a 1-1
To 1-4 are read in parallel (S12). Then, with the former a ₀ ~a ₃ thus read out, the old disk parity PP11~PP41 it has held, from the new a ₀ ~a ₃ Metropolitan,
The values of the new parities PP11 to PP41 to be written to the sector 4 of the disk devices 1-1 to 1-4 are calculated, and the calculated values are held as new old in-disk parity (S13).
Then, write the new a ₀ ~a ₃ to write the sectors 0 new parity P0 which is the parity in parallel (S1
4).

Next, it is determined whether or not the data to be written is completed (S15). In this case, since the data b remains, the process proceeds to step S17. In step S17, the data b to be written next is the same as in step S10.
Then, the parity sector in the disk to be changed is obtained, and it is checked whether the sector is the same as the previous one. In this case, since the sector of the parity in the disk to be changed is the same in the sector 4, the process returns to the step S12. Then, data b is written in the same manner as data a. In this case, the write data is completed (S15), and the held parity PP11-PP41 in the old disk and the parity Pn are stored in the sector 4 of each disk device 1-1-1-4,2.
Are written in parallel (S16).

In the case of continuous access spanning sectors of parity in the disk, as in the case where data a to e from sector 0 to sector 5 in the disk array 1 in FIG. When the writing of the data d, which is the sector immediately before the sector of the internal parity, is completed, it is detected that the parity sector in the disk of the data e to be written next is different from the previous sector. ~ PP
After 41 is once written back, the process returns to step S11 and the above-described processing is repeated.

(2) Reading in the first mode This operation is the same as in the embodiment of FIG.

(3) Reading in the second mode This operation is basically the same as that of the embodiment of FIG. 1, but since there are a plurality of in-disk parities in each of the disk devices 1-1 to 1-4, The parity in the disk corresponding to the sector to be read is read and checked. For example, when reading data a, the second reading unit 44
Data from the sector 0 by accessing the disk device 1-1 to 1-4 and 2 simultaneously in parallel a ₀ ~a ₃ and parity P1
And performs a parity check. If there is no error, the disk devices 1-1 to 1-4 are simultaneously accessed in parallel to read data up to sector 4 and perform a parity check. The parity check performed at this time is a check closed for each of the disk devices 1-1 to 1-4.
In the disk device 1-1, if the parity generated by the data from sector 0 to sector 3 matches the parity PP11 of sector 4, no error occurs. The same applies to the other disk devices 1-2 to 1-4. If even there is no error in any of the disk device, and outputs to the computer 3 from the data a ₀ ~a ₃ read previously assembled data a. The disk devices 1-1 to 1-
Parities PP11 to PP4 read from sector 4 of No. 4
It is also possible to check until the parity generated from 1 matches the parity P4 recorded in the sector 4 of the disk device 2.

[0032]

As described above, the disk array device of the present invention holds the parity of the data recorded in each sector in the data storage disk device separately from the conventional parity storage disk device in RAID3. Therefore, although the overhead at the time of data writing increases, it is possible to ensure extremely high reliability by performing double checking using two types of parities.

According to the configuration of the second aspect, high-speed reading and highly reliable reading can be arbitrarily selected.

Furthermore, in the configuration according to the fourth aspect, the number of sectors to be read at the time of the parity check by the parity in the disk is smaller than the configuration according to the third aspect.
The accompanying overhead can be reduced.

[Brief description of the drawings]

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

FIG. 2 is a diagram illustrating an example of write data.

FIG. 3 is a flowchart illustrating a processing example of a writing unit according to an embodiment of the present invention.

FIG. 4 is a configuration diagram of another embodiment of the disk array device of the present invention.

FIG. 5 is a flowchart illustrating a processing example of a writing unit according to another embodiment of the present invention.

FIG. 6 is a diagram illustrating a configuration example of a conventional disk array device of RAID3.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Disk array 1-1 to 1-4 ... 1st disk device 2 ... 2nd disk device 3 ... Computer 4 ... Disk array control device 41 ... Disk access unit 42 ... Writing unit 43 ... 1st reading unit 44 ... Second readout unit 45. Interface unit

Claims (4)

[Claims] In a disk array device for performing data input / output processing by accessing a plurality of disk devices in parallel, a parity of data recorded in a sector of the own disk is stored in a specific sector of the own disk. A plurality of first disk devices that record data as sectors, and a second disk device that uses each sector for storing parity of data recorded in the sector at the same position of the plurality of first disk devices. A disk array device comprising: a disk array; and a disk array controller that performs data and parity input / output processing for the disk array. 2. The disk array control device divides data into sector units, and divides the data together with its parity into a first data.
A writing unit for writing data in parallel to the same sector of the second disk device and rewriting the parity in the disk of the first disk device to be changed by the writing; And accessing the same sector of the second disk device in parallel and performing a parity check based on the data of the sector read from the plurality of first disk devices and the parity of the same sector read from the second disk device. A first read unit and, when a data read request specifying the second mode is made, accesses the same sector of the first and second disk devices in parallel, and writes the data of the sector read from the plurality of first disk devices and And a parity check based on the parity of the same sector read from the second disk device. ,
2. The disk array according to claim 1, further comprising: a second read unit that reads out parity in the disk and data of a sector from which the parity is generated for each first disk device and performs a parity check based on the data. apparatus. 3. The disk array device according to claim 2, wherein each first disk device is provided with one sector for storing parity in the disk. 4. A system according to claim 2, wherein each first disk device is provided with one sector for storing parity in the disk for every several consecutive sectors.
The disk array device according to the above.