JP2006003971A - Redundant drive group, and writing method and read-out method using the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To restore data not only when a part of data drives become abnormal but also when all of data drives become abnormal. <P>SOLUTION: This redundant drive group is provided with three or more (odd number of) data drives and parity drives of the same number of the data drives. One parity drive among the odd number of the parity drives is prepared corresponding to all data drives. Each of the parity drives except the one parity drive among the odd number of parity drives is prepared corresponding to the data drives which adjoin mutually. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a redundant drive group configured by adding a parity drive to a data drive, and a writing method and a reading method using the redundant drive group.

  In a conventional RAID (Redundant Array of Inexpensive Disks), data is lost when a plurality of drives fail except for RAID1. In RAID1, data is lost if both drives in a mirrored relationship fail.

In order to solve such RAID problems, a plurality of data drives are arranged in a ring, and the same number of parity drives as the data drives are prepared so that each parity drive corresponds to two adjacent data drives. A redundant drive group has been invented (see Patent Document 1).
JP 2000-207136 A

  However, in the invention described in Patent Document 1, data cannot be restored if all data drives become abnormal (see Patent Document 1, paragraph 27).

  Therefore, the present invention provides a redundant drive group that enables data restoration not only when some data drives become abnormal but also when all data drives become abnormal, and this It is an object to provide a writing method and a reading method using the.

  According to the present invention, an odd number of three or more data drives and the same number of parity drives as the data drive are provided, and one of the odd number of parity drives includes all the data drives. Each of the parity drives other than the one parity drive among the odd number of parity drives is provided corresponding to a data drive adjacent to each other. A group is provided.

  According to the present invention, in the redundant drive group, one parity drive provided corresponding to all the data drives among the odd number of parity drives is all the odd number of data drives. If it is normal, the parity calculated based on the data stored in all the data drives is stored. If all of the odd number of data drives are not normal, all the data drives are stored. If parity is normal, the parity calculated based on the data that would have been stored in these data drives is stored, and the parity drives other than the one parity drive among the odd number of parity drives are stored. Each of the data stored in the two corresponding data drives when the two corresponding data drives are normal. The calculated parity is stored, and when one of the two corresponding data drives is not normal, the parity is stored in the normal data drive of the two corresponding data drives A redundant drive group is provided that stores parity calculated based on data and data that would have been stored in the data drive if the data drive and the abnormal data drive were normal.

  Furthermore, according to the present invention, in the writing method using the redundant drive group, the data that has been written to the target data drive, the data to be written to the target data drive, the target data drive and the target Based on the parity previously written to the parity drive corresponding to the data drive adjacent to the data drive, the parity to be written to the parity drive corresponding to the target data drive and the data drive adjacent to the target drive is calculated. Thus, a writing method is provided, wherein the calculated parity is written to a parity drive corresponding to the target data drive and a data drive adjacent to the target drive.

  Furthermore, according to the present invention, in the write method using the redundant drive group, the data that has been written to the target data drive so far, the data to be written to the target data drive, and the parity corresponding to all the data drives. Based on the parity that has been written to the drive so far, the parity to be written to the parity drive corresponding to all the data drives is calculated, and the calculated parity is converted to the parity drive corresponding to all the data drives. A writing method characterized by writing is provided.

  Furthermore, according to the present invention, in the writing method using the redundant drive group, the data that has been written to the target data drive up to now is stored in the data drive adjacent to the target data drive, and There is provided a writing method characterized by performing restoration based on parity stored in a parity drive corresponding to the target data drive and the adjacent data drive.

  Further, according to the present invention, in the writing method using the redundant drive group, the data written so far in the target data drive is stored in all the data drives other than the target data drive. There is provided a writing method characterized in that data is restored based on parity stored in a parity drive corresponding to all drives.

  Furthermore, according to the present invention, in the writing method using the redundant drive group described above, the parity data stored in the parity drives corresponding to all the drives, Provided is a writing method characterized by restoring data of a target drive based on parity stored in some parity drives among parity drives corresponding to any mutually adjacent data drives .

  Further, according to the present invention, in the above writing method, the parity drive corresponding to any one of the mutually adjacent data drives is the i-th data block and the (i + 1) -th parity drive. When the parity block for the data block is represented as a parity block i · (i + 1), when the target data drive is an even-numbered 2m data drive (m is a natural number), the parity block 1 to 2 to the parity block (2m− 1) · 2m and parity block 2m · (2m + 1) to parity block (N−1) · N, and the target data drive is an odd-numbered (2m + 1) (m is a natural number) data block, parity Blocks 1 and 2 to parity block (2m−1) · 2m and parity block 2 (m + ) · (2 (m + 1) +1) to writing method which is a parity block (N-1) · N is provided.

  Furthermore, according to the present invention, in the read method using the redundant drive group, the data stored in the data drive adjacent to the target data drive and the parity corresponding to the target data drive and the adjacent data drive are stored. There is provided a read method characterized in that the data of the target drive is restored based on the parity stored in the drive.

  Furthermore, according to the present invention, in the reading method using the redundant drive group described above, the data stored in all data drives other than the target data drive and the parity drive corresponding to all the drives are stored. There is provided a read method characterized by restoring the data of the target drive based on parity.

  Furthermore, according to the present invention, in the read method using the redundant drive group, the parity stored in the parity drive corresponding to all the drives and the parity drive corresponding to any one of the adjacent data drives There is provided a read method characterized in that the data of the target drive is restored based on the parity stored in some of the parity drives.

  Further, according to the present invention, in the above read method, a part of parity drives corresponding to any one of adjacent data drives is the i-th data block and the (i + 1) -th data drive. When the parity block for the data block is represented as a parity block i · (i + 1), when the target data drive is an even-numbered 2m data drive (m is a natural number), the parity block 1 to 2 to the parity block (2m− 1) · 2m and parity block 2m · (2m + 1) to parity block (N−1) · N, and the target data drive is an odd-numbered (2m + 1) (m is a natural number) data block, parity Blocks 1 and 2 to parity block (2m−1) · 2m and parity block 2 (m + ) · (2 (m + 1) +1) to reading method which is a parity block (N-1) · N is provided.

  According to the present invention, it is possible to restore data not only when some data drives become abnormal, but also when all data drives become abnormal.

  The best mode for carrying out the present invention will be described below in detail with reference to the drawings.

  The present invention provides a computer system having a storage device, in addition to a data drive, a parity drive having parity blocks of all related data blocks and a parity drive having parity blocks related to each pair (pair) of data blocks. It is characterized by a redundant configuration. At the same time, it is possible to recover data when any two drive failures occur, and it is possible to recover data from all data drives even when all data drives have failed.

[1 Parity information]
[1.1 Relationship between data and parity]
FIG. 1 is a relational diagram of data blocks and parity blocks when N data drives are used. However, the total number N of data drives is an odd number of 3 or more. Each parity block in the parity drive holds parity information related to the data block associated with the solid line.

  Parity information satisfying the following relational expression is held in the parity block in each parity drive. When a data block is updated, it is necessary to update the associated parity block as well as the contents of the data block in order to maintain the relevance.

Parity block n-1 ・ n
= Data block n-1 XOR Data block n
<Formula 1>
(Where n = 2, 3,..., N)
Parity blocks 1 to N
= Data block 1 XOR
Data block 2 XOR
...
Data block N-1 XOR
Data block N
<Formula 2>
However, the parity block n-1 · n indicates that the parity block is related to the data block n-1 and the data block n, and the parity blocks 1 to N are related to all of the data blocks 1 to N. XOR indicates a parity block, and XOR indicates exclusive OR. The calculation of exclusive logic is performed using a logic gate as shown in FIG. 2 (a) shows a logic gate in the case of two inputs, FIG. 2 (b) shows a logic gate in the case of three inputs, and FIGS. 2 (c) and 2 (d) show N input inputs. Shows the logic gates of the case. When a CPU is used, a logic gate as shown in FIG. 2 is realized by an ALU provided inside or outside the CPU. In the case of 3 inputs or more, a register is also used, and the output y of FIG. 2A is temporarily stored in the register, and then given to the logic gate as input x1, and input to the other end of the logic gate. You may make it take an exclusive OR with data.

[1.2 Read / write operation overview]
An outline of the read / write operation is shown in the flowcharts of FIGS. Each step described with reference to these flowcharts can be realized by hardware, but can also be realized by a computer incorporating a CPU reading and executing a program.

  When realized by hardware, for example, read / write is performed by a read / write device as shown in FIG.

  Referring to FIG. 5, the read / write device includes a data reading unit 251, a parity reading unit 252, a read / write determining unit 253, a data drive availability determining unit 254, a data reproducing unit 255, a parity generating unit 256, a data writing unit. Unit 257 and a parity writing unit 258.

  The read / write determination is performed by the read / write determination unit 253 determining a write command or a read command. The read command and the write command include identification information for identifying the access destination data drive and an address for identifying the access destination address, and the write command includes data to be written.

[2 Read processing]
[2.1 When Data Drive of Target Data Block is Available (YES at 202 → 208)]
In this case, the data reading unit 251 reads from the target data drive (209).

[2.2 When the drive with the target data block is not available (NO in 202 → 208)-No. 1]
Parity block information related to the target data block (read out by the parity reading unit 252) and data block information related to the parity block on the other side (read out by the data reading unit 251) Is used to restore the information in the target data block by the data reproduction unit 255 (215). The restoration of the data block uses one of the following formulas that can be calculated by the target data block and the accessible drive.

Data block 1
= Data block 2 XOR Parity block 1
<Formula 3-1-1>
Data block 1
= Data block 2 XOR
Data block 3 XOR
...
XOR
Data block N XOR
Parity blocks 1 to N
<Formula 3-1-2>
According to <Equation 3-1-1>, it is possible to restore the data of the target block 1 based on the adjacent data block and the target data block 1 and the parity block related to the adjacent data block.

  According to <Equation 3-1-2>, the data of the target data block 1 can be restored based on all the data blocks other than the target data block 1 and the parity blocks 1 to N.

Data block N
= Data block N-1 XOR
Parity block N-1 ・ N
<Formula 3-2-1>
Data block N
= Data block 1 XOR
Data block 2 XOR
...
XOR
Data block N-1 XOR
Parity blocks 1 to N
<Formula 3-2-2>
According to <Formula 3-2-1>, it is possible to restore the data of the target block N based on the adjacent data block, the target data block N, and the parity block related to the adjacent data block.

  According to <Formula 3-2-2>, the data of the target data block N can be restored based on all the data blocks other than the target data block N and the parity blocks 1 to N.

Data block n
= Data block n + 1 XOR
Parity block n · n + 1
(N = 2, 3, ..., N-1)
<Formula 3-3-1>
Data block n
= Data block n-1 XOR
Parity block n-1 ・ n
(N = 2, 3, ..., N-1)
<Formula 3-3-2>
Data block n
= Data block 1 XOR
...
XOR
Data block n-1 XOR
Data block n + 1 XOR
...
XOR
Data block N XOR
Parity blocks 1 to N
(N = 2, 3, ..., N-1)
<Formula 3-3-3>
According to <Equation 3-3-1> and <Equation 3-3-2>, the data of the target block N is restored based on the adjacent data block, the target data block N, and the parity block related to the adjacent data block. It becomes possible to do.

  According to <Expression 3-3-3>, the data of the target data block N can be restored based on all the data blocks other than the target data block N and the parity blocks 1 to N.

[2.3 When all data drives have failed (NO at 202 → 208) to 2]
If all the parity drives are available, the information of all data drives can be restored by the following formula (215).

[2.3.1 Reading data block of even number (2m)]
Data block a
= Parity block 1 to N XOR
Parity block 1 and 2 XOR
...
Parity block (2m-1) · 2m XOR
Parity block 2m · (2m + 1) XOR
...
Parity block (N-1) ・ N
<Formula 4-1>
[2.3.2 Reading data block of odd number (2m + 1)]
Data block b
= Parity block 1 to N XOR
Parity block 1 and 2 XOR
...
Parity block (2m-1) · 2m XOR
Parity block 2 (m + 1) · (2 (m + 1) +1) XOR
...
Parity block (N-1) ・ N
<Formula 4-2>
[3 Write processing]
[3.1 When Data Drive of Target Data Block n is Available (YES at 202 → 203)]
The parity block associated with the data block n is recalculated based on the following equation after the parity information associated with the data block n is erased (204).

  The data writing unit 257 and the parity writing unit 258 write the new parity information and write data generated by <Expression 5> <Expression 6> to the parity block and the data block (205, 206).

New parity blocks 1 to N
= Current parity block 1 to N XOR
Current data block n XOR
New data block n
(N = 1, 2, 3,..., N)
<Formula 5>
New parity block n-1 ・ n
= Current parity block n-1 · n XOR
Current data block n XOR
New data block n
(N = 2, 3, ..., N)
<Formula 6>
New parity block n · n + 1
= Current parity block n · n + 1 XOR
Current data block n XOR
New data block n
(N = 1, 2, 3, ..., N-1)
<Formula 7>
The current parity is performed by the parity reading unit 252. Reading of the current data block is performed by the data reading unit 251. Calculations of <Expression 5> to <Expression 7> are performed by the parity generation unit 256.

[3.2 When the drive of the target data block is not available (NO in 202 → 203)
First, the data reproducing unit 255 reproduces old information of the target data block (211). For this reproduction, the method of step 215 is used. Specifically, the method described in [2.2] or [2.3] is used.

  Next, the parity generation unit 256 calculates new information of the related parity block from the reproduced information, the current information of the related parity block, and the information to be newly written, and writes it by the parity writing unit 258. (213). Specifically, three types of related parities are calculated using the method described in [3.1] above, and these are written in the corresponding parity drives.

  The embodiment of the present invention is as shown in FIG. 6 when N = 5.

  Here, an example of reading at the time of failure when two drives have failed and when all of the data drives have failed is shown.

[1. Two drives fail]
[1.1 When both failures are parity drives]
Since all data drives are available, data can be read.

[1.2 When both failures are data drives]
[1.2.1 When data drive 1 (301) and data drive 5 (305) fail]
The data block 1 (311) in the data drive 1 (301) can be reproduced from <Equation 3-1-1> from the parity blocks 1 and 2 (316) and the data block 2 (312).

  The data block 5 (315) in the data drive 5 (305) can be reproduced from <Equation 3-2-1> from the parity blocks 4 and 5 (319) and the data block 4 (314).

[1.2.2 When data drive 1 (301) and data drive 2 (302) fail]
The data block 2 (312) in the data drive 2 (302) can be reproduced from the data block 3 (313) and the parity blocks 2 and 3 (317) according to <Equation 3-3-1>.

  The data block 1 (311) in the data drive 1 (301) can be reproduced from <Expression 3-3-2> from the data block 2 (312) and the parity blocks 1 and 2 (316) reproduced as described above.

  The data block 1 (311) in the data drive 1 (301) can also be reproduced from the parities 1 to 5, the parities 1 and 2, and the parities 3 and 4.

[1.2.3 (When data drive 2 (302) and data drive 3 (303) fail)
The data block 2 (312) can be reproduced from the parity blocks 1 and 2 (316) and the data block 1 (311) according to <Expression 3-3-2, n = 2>.

  The data block 3 (313) can be reproduced from the parity blocks 3 and 4 (318) and the data block 4 (314) according to <Equation 3-3-1, n = 3>.

[1.3 When two failures are distributed to the parity drive and data drive]
[1.3.1 When data drive 1 (301) and parity drives 1 to 5 (300) fail]
The restoration of the data block 1 (311) can be reproduced from the parity blocks 1 and 2 (316) and the data block 2 (312) according to <Equation 3-1-1>.

[1.3.2 When data drive 1 (301) and parity drives 1 and 2 (316) fail]
The data block 1 (311) can be recovered from the data blocks 2 (312) to 5 (315) and the parity blocks 1 to 5 (310) according to <Equation 3-1-2>.

[1.3.3 When Data Drive 2 (302) and Parity Drives 1-5 (300) Drive Fail]
The data block 2 (312) can be restored from the parity blocks 1 and 2 (316) and the data block 1 (311) according to <Expression 3-3-2, n = 2>.

[1.3.4 When data drive 2 (302) and parity drives 1 and 2 (306) fail]
Data block 2 (312) can be restored from parity blocks 2 and 3 (317) and data block 3 (313) <Equation 3-3-1, n = 2> [2. When all data drives fail]
Data blocks can be reproduced as follows.

  Data block 1 (311) is reproducible from <Equation 4-1> from parity blocks 2 and 3 (317), parity blocks 4 and 5 (319), and parity blocks 1 to 5 (310).

  Data block 2 (312) includes parity blocks 1 and 2 (316), parity blocks 2 and 3 (317), parity blocks 4 and 5 (319), and parity blocks 1 to 5 (310) according to <Formula 4-2>. It is reproducible.

  Data block 3 (313) is reproducible from <Equation 4-1> from parity blocks 1 and 2 (316), parity blocks 4 and 5 (319), and parity blocks 1 to 5 (310).

  Data block 4 (314) is reproduced from parity blocks 1 and 2 (316), parity blocks 3 and 4 (318), parity blocks 4 and 5 (319), and parity blocks 1 to 5 (310) <Equation 4-2>. Is possible.

The data block 5 (315) can be reproduced from the parity blocks 1 and 2 (316), the parity blocks 3 and 4 (318), and the parity blocks 1 to 5 (310) <Equation 4-1>.
A list of these relationships is shown in FIG. FIG. 8 shows a similar relationship list when N is 3, FIG. 9 shows a similar relationship list when N is 7. FIG. 9 shows a similar relationship list when N is 9. Is shown in FIG. When N is 1, it is obvious that the data block 1 can be reproduced using the parity block 1.

  FIG. 11 summarizes the list for N = 1, N = 3, N = 5, N = 7, and N = 9. When N = 1, a parity block indicated by a circle in an area surrounded by an inverted L-shaped frame indicated by reference numeral 301 is used, and when N = 3, an inverted L-character shape indicated by reference numeral 303 is used. The parity block indicated by ◯ in the area surrounded by the frame is used, and when N = 5, the parity block indicated by ◯ in the area surrounded by the inverted L-shaped frame indicated by reference numeral 305 is used. When N = 7, the parity block indicated by ◯ in the region surrounded by the inverted L-shaped frame indicated by reference numeral 307 is used. When N = 9, the inverse L indicated by reference numeral 309 is used. A parity block indicated by a circle in an area surrounded by a character frame is used.

  As is clear from FIG. 11, in order to increase the value of N from 2m + 1 (m = 0, 1, 2,...) To 2 (m + 1) +1, it is used when the value of N is 2m + 1. The parity block is used as it is, and for the extended area, which parity block is used is determined according to a predetermined rule (a constant rule that does not depend on the value of m).

  That is, when the number of data blocks increases from 2m + 1 to 2 (m + 1) +1, in order to reproduce 1 to 2m + 1 data blocks of 1 to 2 (m + 1) +1 data blocks, N All the used parity when the value is 2m + 1 is used (however, the parity for all the data blocks is changed from the parity block for the data blocks 1 to 2m + 1 to the parity block for the data blocks 1 to 2 (m + 1) +1). In addition, a parity block for data block 2 (m + 1) and data block 2 (m + 1) +1 is additionally used.

  In order to reproduce the data block 2 (m + 1), all the parity blocks used for reproducing the data block 2m + 1 are used, and the parity block and the data block 2 (for the data block 2m + 1 and the data block 2 (m + 1)) are also used. Use parity blocks for m + 1) and data block 2 (m + 1) +1.

  In order to reproduce the data block 2 (m + 1) +1, all the parity blocks used for reproducing the data block 2m + 1 are used. Furthermore, the parity block and the data block 2 for the data block 2m + 1 and the data block 2 (m + 1) are used. Use (m + 1).

  <Formula 4-1> and <Formula 4-2> are clear from FIGS.

  When the value of N is an even number, the values of the parity blocks 1 to N are, for example, when the values of all the data blocks are 1, even when the values of all the data blocks are 0. Even if it exists, since it becomes 0 and cannot be distinguished, the present invention cannot be used when the value of N is an even number.

  The present invention is error-resistant and can be used for reliable data writing and reading.

It is a conceptual diagram which shows the structure of the redundant drive group by embodiment of this invention. It is a circuit diagram which shows the structure of the parity generation part by embodiment of this invention. It is a 1st flowchart which shows the write method and read method using the redundant drive group by embodiment of this invention. It is a 2nd flowchart which shows the write method and read method using the redundant drive group by embodiment of this invention. It is a block diagram which shows the structure of the writing / reading apparatus by embodiment of this invention. It is a conceptual diagram which shows the structure of the redundant drive group by Example 1 of this invention. 10 is a table showing parity used for reproducing data when all data blocks become abnormal when the number of data drives is five. 10 is a table showing parity used for reproducing data when all data blocks become abnormal when the number of data drives is three. 10 is a table showing parity used for reproducing data when all data blocks become abnormal when the number of data drives is seven. 10 is a table showing parity used for reproducing data when all data blocks become abnormal when the number of data drives is nine. 10 is a table showing parity used for reproducing data when all data blocks become abnormal when the number of data drives is 1, 3, 5, 7, or 9.

Explanation of symbols

251 Data reading unit 252 Parity reading unit 253 Read / write determining unit 254 Data drive availability determining unit 255 Data reproducing unit 256 Parity generating unit 257 Data writing unit 258 Parity writing unit

Claims (14)

An odd number of three or more data drives;
The same number of parity drives as the data drive;
With
One parity drive among the odd number of parity drives is provided corresponding to all data drives,
Each of the parity drives other than the one parity drive among the odd number of parity drives is provided corresponding to a data drive adjacent to each other. In the redundant drive group according to claim 1,
One parity drive provided corresponding to all data drives of the odd number of parity drives is stored in all data drives when the odd number of data drives are all normal. The parity calculated based on the data stored in the data is stored, and in the case other than the case where all the odd number of data drives are normal, if all the data drives are normal, they are stored in these data drives. Store the parity calculated based on the data that would be
Of the odd number of parity drives, each of the parity drives other than the one parity drive is stored in the corresponding two data drives when the corresponding two data drives are both normal. The parity calculated based on the data is stored, and if one of the two corresponding data drives is not normal, the parity is stored in the normal data drive of the two corresponding data drives. A redundant drive group characterized by storing parity calculated based on data that would be stored in this data drive if the data being stored and the data drive that is not normal are normal. In the writing method using the redundant drive group according to claim 1 or 2,
The data that has been written to the target data drive, the data to be written to the target data drive, and the parity that has been written to the parity drive corresponding to the target data drive and the data drive adjacent to the target data drive. Based on the target data drive and a parity drive corresponding to the data drive adjacent to the target drive, a parity to be written is calculated, and the calculated parity is used as the data drive adjacent to the target data drive and the target drive. And writing to a parity drive corresponding to the above. In the writing method using the redundant drive group according to claim 1 or 2,
Corresponds to all data drives based on the data that has been written to the target data drive, the data to be written to the target data drive, and the parity that has been written to the parity drive corresponding to all data drives. A writing method comprising: calculating a parity to be written to a parity drive to be written, and writing the calculated parity to a parity drive corresponding to all the data drives. In the writing method using the redundant drive group according to claim 3 or 4,
Data that has been written to the target data drive so far is stored in data stored in a data drive adjacent to the target data drive and in a parity drive corresponding to the target data drive and the adjacent data drive. A writing method characterized by performing restoration based on parity. In the writing method using the redundant drive group according to claim 3 or 4,
The data written to the target data drive so far is restored based on the data stored in all data drives other than the target data drive and the parity stored in the parity drive corresponding to all the drives. A writing method characterized by the above. In the writing method using the redundant drive group according to claim 3 or 4,
The data that has been written to the target data drive up to now, the parity stored in the parity drive corresponding to all the drives, and the parity of some of the parity drives corresponding to any mutually adjacent data drives A writing method, comprising: restoring data of a target drive based on parity stored in the drive. The writing method according to claim 7,
Some of the parity drives corresponding to any mutually adjacent data drives are:
When the parity block for the i-th data block and the (i + 1) -th data block is represented as a parity block i · (i + 1),
When the target data drive is an even-numbered 2m (m is a natural number) data drive, parity blocks 1 to 2 to parity blocks (2m-1) to 2m and parity blocks 2m to (2m + 1) to parity blocks (N -1) · N,
When the target data drive is an odd-numbered (2m + 1) (m is a natural number) data block, parity block 1 to 2 to parity block (2m−1) · 2m and parity block 2 (m + 1) · (2 ( m + 1) +1) to parity block (N−1) · N. In the reading method using the redundant drive group according to claim 1 or 2,
Restoring the data of the target drive based on the data stored in the data drive adjacent to the target data drive and the parity stored in the parity drive corresponding to the target data drive and the adjacent data drive; A characteristic readout method. In the reading method using the redundant drive group according to claim 1 or 2,
A read method comprising: restoring data of the target drive based on data stored in all data drives other than the target data drive and parity stored in parity drives corresponding to all the drives. In the reading method using the redundant drive group according to claim 1 or 2,
Based on the parity stored in the parity drive corresponding to all the drives and the parity stored in some parity drives among the parity drives corresponding to any of the data drives adjacent to each other, A reading method characterized by restoring data. The reading method according to claim 11,
Some of the parity drives corresponding to any mutually adjacent data drives are:
When the parity block for the i-th data block and the (i + 1) -th data block is represented as a parity block i · (i + 1),
When the target data drive is an even-numbered 2m (m is a natural number) data drive, parity blocks 1 to 2 to parity blocks (2m-1) to 2m and parity blocks 2m to (2m + 1) to parity blocks (N -1) · N,
When the target data drive is an odd-numbered (2m + 1) (m is a natural number) data block, parity block 1 to 2 to parity block (2m−1) · 2m and parity block 2 (m + 1) · (2 ( m + 1) +1) to parity block (N−1) · N.   The program for making a computer perform the writing method of any one of Claim 3 thru | or 8.   A program for causing a computer to perform the reading method according to any one of claims 9 to 12.

Images