JPH04321123A - Disk array controller - Google Patents

Abstract

PURPOSE:To improve the reliability of a compact computer system using plural disk array devices by attaining a 2-dimensional parity check system including the horizontal and vertical series. CONSTITUTION:If one of disk DD0-DD5 of a horizontal series is faulty and the reproduction of data is impossible, the data recovery processing is carried out based on the horizontal parity data to which the faulty disk drive belongs. Meanwhile the data recovery processing is carried out with use of the vertical parity data stored in the disk drives P0-P5 exclusive for the parity data if one of disk drives of a vertical series has a fault and the reproduction of data is impossible.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array control device that controls a plurality of disk drives and has a data recovery function in the event of a data error.

0002

2. Description of the Related Art Conventionally, a disk array device has a plurality of disk drives (DD), and each disk drive is configured to be controlled by a common controller (DAC). The controller is connected to a host computer via a host bus, and accesses each disk drive in response to instructions from the host computer.

[0003] In disk array devices, a parity check method is adopted in order to improve the reliability of the device. Specifically, a method is used in which each write data of the same byte written to the same physical sector of each disk drive is subjected to an exclusive OR operation, and parity data is generated and stored based on the result of this operation. ing.

[0004] When a failure occurs in a disk drive, the controller uses stored parity data to restore data that cannot be reproduced due to the failure. Parity data is stored in a disk drive dedicated to parity, or distributed and stored in each disk drive. Normally, a method of storing parity on a dedicated disk drive results in a decrease in access speed, so a method of storing data in a distributed manner is adopted.

By the way, a computer system configuration using a plurality of disk array devices is conceivable. In such a system, an array is constructed consisting of a horizontal (horizontal) array of disk drives belonging to each disk array device and a vertical (vertical) array of disk drives that extends beyond each disk array device. .

[0006] In such a system, the number of disk drives is large, so the failure rate of multiple drives at the same time is high, but the one-dimensional parity check method using only horizontal series parity data as described above , data recovery processing is impossible.

[0007]

[Problem to be Solved by the Invention] In a system using a plurality of disk array devices, it is difficult to maintain high reliability with a one-dimensional parity check method that uses only horizontal series parity data for each disk array device. is not possible. For this reason, it is conceivable to adopt a two-dimensional parity check method. However, although the two-dimensional parity check method has been implemented in large-scale systems, it has not been implemented in small-sized computer systems.

An object of the present invention is to realize a two-dimensional parity check method that utilizes parity data in horizontal and vertical sequences in a small computer system using a plurality of disk array devices, thereby improving reliability. An object of the present invention is to provide a disk array control device that can realize a system with high performance.

[0009]

[Means for Solving the Problems] The present invention regards a plurality of disk drives in a horizontal series as one unit, and provides a main disk array control means, a sub-disk array control means, and a parity disk array provided for each unit. This is a disk array control device consisting of a control means. The main disk array control means and the sub-disk array control means communicate with each other, and each uses horizontal parity data generated based on each write data recorded in the same predetermined storage location of each disk drive in the horizontal series. Perform data recovery processing.

In the parity disk array control means, the main and sub disk array control means communicate with each other, and data is recorded at the same predetermined storage location of each disk drive in a longitudinal series beyond each disk array control means. A parity data dedicated disk drive that stores vertical parity data generated based on each write data is controlled.

[0011]

[Operation] According to the present invention, when one disk drive among the disk drives in the horizontal series has a failure and data cannot be reproduced, data is not reproduced based on the horizontal parity data to which the failed disk drive belongs. , data repair processing is performed. On the other hand, if one of the disk drives in the vertical series fails and data cannot be played back, the vertical parity data stored in the parity data-dedicated disk drive is used to A repair process is performed. As a result, even if multiple disk drives fail at the same time, data recovery processing can be performed based on the horizontal parity data and the vertical parity data.

0012

DESCRIPTION OF THE PREFERRED EMBODIMENTS Examples of the present invention will be described below with reference to the drawings.

FIG. 1 shows a disk array according to the same embodiment.
FIG. 2 is a block diagram showing main parts of the system. This system consists of a host computer 1, a main disk array device, a plurality of sub-disk array devices, and a disk array device exclusively for parity data. The host computer 1 sends commands and sends/receives data to each disk array device via the host bus 1a.

[0014] The main disk array device is
It consists of a controller (main DAC) 2 and a main disk drive group 6. The main disk drive group 6 includes, for example, six disk drives DD0 to DD5. The controller 2 controls each of the horizontal series of disk drives DD0 to DD5 through the local bus 2a.

Each sub-disk array device consists of a sub-controller (DAC) 3 and a sub-disk drive group 7. The sub-disk drive group 7 has the same configuration as the main, and includes six disk drives DD0 to
Consists of DD5. The controller 3 is connected to the local bus 3a
Through each disk drive DD0~D in the horizontal series
D5 respectively.

The parity data-dedicated disk array device consists of a parity controller (PDC) 4 and a parity disk drive group 8. The parity disk drive group 8 has the same configuration as the main and sub disk drive groups, and has six disk drives P0.
~ Consists of P5. The controller 4 has a local bus 4a.
Through each disk drive P0 to P5 in the horizontal series
control each.

Each of the controllers 2 to 4 has a host bus 1a.
and communicate with each other via a parity bus 5. Each of the controllers 2 to 4 includes a CPU 10, a memory (ROM) 11, a buffer memory 12, and a parity generation circuit 13, as shown in FIG. C
PU10 is a controller main body that executes control operations. The memory 11 is a read-only memory that stores programs for the CPU 10 and various control data. The buffer memory 12 is a memory that temporarily stores data to be transferred to and from the host computer 1. The parity generation circuit 13 is a circuit for generating parity data based on write data of each disk drive.

Furthermore, each controller 2 to 4 has a local bus interface (I) connected to the local bus.
/F) 14, host bus I connected to host bus 1a
/F15 and a parity bus I/F16 connected to the parity bus 5. Next, the operation of this embodiment will be explained.

First, the main DAC 2 and each sub DAC
3 is each disk drive DD0 to DD5 in the horizontal direction.
As shown in FIG. 3(B), logical sectors (0 to 19...) are sequentially allocated to each physical sector (0 to 3...) of each disk drive, and a parity sector for parity data Ph is assigned to each physical sector. will be established.

The parity generation circuit 13 of the main DAC 2 performs an exclusive OR operation on the write data in bytes written to each logical sector belonging to the same physical sector in the horizontally arranged disk drive group 6, and Parity data Ph is generated based on the calculation result. Main D
The CPU 10 of the AC2 stores the generated parity data Ph in the parity sector of the same physical sector. That is, the horizontal parity data Ph corresponding to each disk drive DD0 to DD5 in the horizontal series is
It is distributed and stored in D0 to DD5.

Similarly, in each of the disk drives DD0 to DD5 in the horizontal series belonging to each sub DAC 3, horizontal parity data Ph is generated by the parity generation circuit 13 and stored in a distributed manner in each of the disk drives DD0 to DD5. Ru.

Specifically, as shown in FIG. 3A, for example, in each disk drive DD0 to DD5, 8-bit write data D0 to D5 of the same physical sector are subjected to an exclusive OR operation, and this 8-bit horizontal parity data Ph is generated based on the calculation result.

On the other hand, the PDC 4 generates vertical parity data Pv corresponding to each disk drive in the vertical series based on the write data transferred from the main DAC 2 and sub DAC 3 via the parity bus 5.
Save to disk drives P0 to P5. That is, P.D.C.
The parity generation circuit 13 of No. 4 has the same vertical disk drive (
For example, each write data belonging to the same physical sector of DD0) is subjected to an exclusive OR operation, and based on the result of this operation, parity data Pv is generated and stored in a disk drive (P0) in the same vertical direction. Each light data is
For example, it is write data in bytes written to each logical sector belonging to the same physical sector of the disk drive DD0.

Specifically, as shown in FIG. 4, 8-bit data (D0-1 to D0-n) of the same logical sector of n disk drives (DD0) in the same vertical direction are subjected to exclusive OR. Arithmetic processing is performed, and vertical parity data Pv is generated based on the result of the calculation. This generated vertical parity data Pv is transferred to the same vertical disk drive (
P0).

In such basic operations, the updating process and data restoration process of horizontal parity data Ph and vertical parity data Pv according to the same embodiment will be explained.

As shown in FIG. 5, the host computer 1
When a read command is output from (NO in step S1)
, the main DAC 2 receives the read command via the host bus 1a. The main DAC2 receives the read command,
It is transferred to each sub DAC 3 as a local read command via the parity bus 5 (step S6). Main DAC
2 calculates the address of the target logical sector to be accessed based on the received read command, and outputs a local read command including the address.

[0027] The main DAC 2 and each sub DAC 3 are
Each of the disk drives DD0 to DD5 in the horizontal series is accessed through the local buses 2a and 3a, respectively, and the data to be accessed is read (step S7). The main DAC 2 and each sub DAC 3 store the read data in each buffer memory 12 (step S9).
. The main DAC 2 outputs commands to each sub-DAC 3 via the parity bus 5 in the order of data to be transferred to the host computer 1. As a result, each sub DAC 3 and main DAC 2 connect to the host bus 1a.
The read data stored in each buffer memory 12 is transferred in a predetermined order (step S10).

On the other hand, when a write command is output from the host computer 1 (YES in step S1), the main D
AC2 receives a write command via host bus 1a. The main DAC 2 sends the received write command to each sub DAC 3 as a local write command via the parity bus 5.
and transfer it to the PDC 4 (step S2). Main D
The AC 2 and each sub DAC 3 receive write data (
Only necessary data is received among the new data (step S3).

The main DAC 2 and each sub DAC 3 are
The received new data and old data to be updated are transferred to the PDC 4 via the parity bus 5 (step S4). The old data is stored in each disk drive DD0 in the horizontal direction.
This is data that was recorded at the address where new data is to be written in the DD5.

The PDC 4 stores the vertical parity data Pv corresponding to the old data to be updated in the disk drives P0 to
Read from P5. The PDC 4 generates new vertical parity data Pv based on the old and new data transferred from the main DAC 2 and each sub-DAC 3 and the read old parity data Pv, and stores it in the disk drives P0 to P5 (step S5). As a result, the vertical parity data Pv is also updated in accordance with the data update for each disk drive DD0 to DD5 in the horizontal series.

[0031] Here, in accordance with the data update for each disk drive DD0 to DD5 in the horizontal series, each horizontal parity data Ph is also updated. In this case, the main DAC 2 and each sub DAC 3 generate new horizontal parity data Ph based on the old and new data and the old parity data Ph, and
5 parity sector.

Next, in the process of step S7,
When reading data, if an error occurs due to a disk drive failure (YES in step S8), the main DAC 2 and each sub DAC 3
performs data recovery processing.

That is, each disk drive D in the horizontal series
If there is one failed disk drive among D0 to DD5 (YES in step S11), the corresponding main DA
The C2 or sub DAC 3 executes data recovery processing for the failed disk drive based on the write data of the normal disk drive and the horizontal parity data Ph (step S12). Specifically, for example, disk drive D
When D3 fails, as shown in FIG. 3A, the error data D3 is repaired based on the normal write data D0 to D2, D4, and D5 of the same physical sector and the horizontal parity data Ph. . Each write data D0~
D2, D4, and D5 are data written to the same physical sector of each disk drive DD0 to DD2 and DD4, DD5, respectively.

The main DAC 2 or sub DAC 3 stores the repaired write data in the buffer memory 12 and transfers it as normal read data to the host computer 1 via the host bus 1a (step S14).

If there are multiple failed disk drives (
(NO in step S11), the corresponding main DAC 2 or sub DAC 3 transfers the local read command to the other DAC and PDC 4 via the parity bus 5 (step S15). That is, each disk drive D in the horizontal series
Among D0 to DD5, normal write data and vertical parity data Pv are read from each disk drive in the same vertical direction as the failed disk drive. The corresponding main DAC 2 or sub DAC 3 executes data recovery processing for the failed disk drive based on each write data in the same vertical direction and the vertical parity data Pv (step S16).

Specifically, for example, DAC-1 (main D
Disk drive D belonging to AC2 or sub DAC3)
When D3 fails, as shown in FIG. 4, based on normal write data D3-2 to D3-n and vertical parity data Pv-3 of the same physical sector of each disk drive DD3 in the same vertical direction, , repair the error data D3-1. The vertical parity data Pv-3 is vertical parity data Pv stored in the same vertical disk drive P3.

Here, if there are two failed disk drives, the data repaired using the same vertical write data and vertical parity data Pv is used, and each normal write data in the horizontal series and the horizontal parity The data of the other device is restored using the data Ph (steps S12 to S
14).

In addition, if there are three or more failed disk drives, each write data of each of the disk drives DD0 to DD5 in the same longitudinal series and each longitudinal parity stored in each of the disk drives P0 to P5 are stored as described above. Data P
v, each data on the failed disk drive is repaired.

In this way, by using the horizontal parity data Ph distributed to each of the disk drives DD0 to DD5 in the horizontal series controlled by the main DAC 2 and each sub-DAC 3, a failure belonging to the horizontal series is detected. You can perform data recovery processing on disk drives. Further, by using the vertical parity data Pv corresponding to each disk drive in the vertical series beyond the main DAC 2 and each sub-DAC 3, it is possible to perform data recovery processing for a failed disk drive belonging to the vertical series. Therefore, by using two-dimensional parity data in the horizontal and vertical directions, it is possible to perform data recovery processing even when a plurality of failed disk drives occur at the same time.

[0040]

As described in detail above, according to the present invention, in a small computer system using a plurality of disk array devices, parity data in the horizontal series for each disk array device and vertical series across each disk array device can be A two-dimensional parity check method using direction series parity data can be realized. Therefore, even if a plurality of failed devices occur at the same time, data recovery processing can be reliably realized, and as a result, a highly reliable system can be constructed.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the main parts of a disk array system according to an embodiment of the present invention.

FIG. 2 is a block diagram showing main parts of a disk array control device according to the same embodiment.

FIG. 3 is a conceptual diagram for explaining the operation of the embodiment.

FIG. 4 is a conceptual diagram for explaining the operation of the embodiment.

FIG. 5 is a flowchart for explaining the operation of the embodiment.

FIG. 6 is a flowchart for explaining the operation of the embodiment.

[Explanation of symbols]

1...Host computer, 2...Main controller,
3...Sub controller, 4...Parity controller, 6, 7...Disk drive group, 8...Parity disk drive group.

Claims (3)

[Claims] 1. Controlling a plurality of disk drives, generating horizontal parity data based on each write data recorded in the same predetermined storage location of each of the disk drives, and recording the horizontal parity data in each of the disk drives; main disk array control means having a data recovery function in response to occurrence of a data error in the disk drive based on horizontal parity data; communicating with the main disk array control means to control a plurality of disk drives;
Horizontal parity data is generated based on each write data recorded in a predetermined storage location of each disk drive and recorded in each of the disk drives, and a data error occurs in the disk drive based on the horizontal parity data. a sub-disk array control means having a data recovery function for data recovery, and each of the disk drives connected to both the main disk array control means and the sub-disk array control means and controlled by each of the disk array control means. is a horizontal series, based on each write data recorded in the same predetermined storage location of each disk drive in a vertical series consisting of one disk drive corresponding to each disk array control means. 1. A disk array control device comprising: a parity disk array control means for generating vertical parity data using a plurality of keys, and controlling a parity data-dedicated disk drive for storing the vertical parity data. 2. Controlling a plurality of disk drives, generating horizontal parity data based on each write data recorded in the same predetermined storage location of each of the disk drives, and recording the horizontal parity data in each of the disk drives; main disk array control means having a data recovery function in response to occurrence of a data error in the disk drive based on horizontal parity data; communicating with the main disk array control means to control a plurality of disk drives;
Horizontal parity data is generated based on each write data recorded in a predetermined storage location of each disk drive and recorded in each of the disk drives, and a data error occurs in the disk drive based on the horizontal parity data. a sub-disk array control means having a data recovery function for data recovery, and each of the disk drives connected to both the main disk array control means and the sub-disk array control means and controlled by each of the disk array control means. is a horizontal series, based on each write data recorded in the same predetermined storage location of each disk drive in a vertical series consisting of one disk drive corresponding to each disk array control means. a parity disk array control means for generating vertical parity data in the vertical series and controlling a parity data-dedicated disk drive for storing the vertical parity data; and in the case where a data error occurs in each disk drive in the vertical series 2. A disk array control device comprising: vertical data restoration means for performing data restoration processing based on the vertical parity data corresponding to the vertical series. 3. A disk array device comprising a plurality of disk drives arranged in a horizontal direction and a plurality of disk drives arranged in a vertical direction, wherein the information is recorded at the same predetermined storage location of each disk drive in the horizontal direction. horizontal parity data generating means for generating horizontal parity data based on each write data and recording it in each of the disk drives; horizontal data restoration means for performing data restoration processing in response to data errors occurring in disk drives in the directional series; and vertical parity based on each write data recorded in the same predetermined storage location of each disk drive in the vertical series. vertical parity data generation means for generating data and recording it on a disk drive dedicated to parity data; and data of the vertical series disk drives based on the vertical parity data generated by the vertical parity data generation means. What is claimed is: 1. A disk array control device comprising: vertical data restoration means for performing data restoration processing in response to occurrence of an error.