JP2000276308A - Disk subsystem and data restoration method in disk subsystem - Google Patents

Abstract

An object of the present invention is to make a disk subsystem unusable because data necessary for restoring failed data cannot be read from a magnetic disk device in which no failure has occurred. It is an object of the present invention to provide a disk subsystem that prevents the occurrence of a disk subsystem. In a disk subsystem configured with RAID, when restoring data of a failed magnetic disk device to the spare magnetic disk device, data readable from the failed magnetic disk device is:
The data that cannot be read from the failed magnetic disk device is stored in the spare magnetic disk device, and the data that cannot be read from the failed magnetic disk device is one of the plurality of magnetic disk devices that constitute the RAID. And data restoring means for restoring the data stored in the spare magnetic disk drive by operation and storing the restored data in the spare magnetic disk device.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement in a disk subsystem using a RAID technology used as an external storage device requiring high reliability in a server computer such as a PC server. In particular, the present invention relates to a disk subsystem using RAID technology for further increasing reliability.

[0002]

2. Description of the Related Art RAID (Redundant Arr.)
ays of Inexpensive Disks)
The technology is described in David A. Patternson, Gar
thA. Gribson, Randy H .; Kats, published in 1987, "A Case fo
r Redundant Arrays of Ine
"xpensive Disks".

[0003] The RAID technology is generally classified into level 1 to level 5 and is a P technology that requires high reliability.
It is used for a disk subsystem of a server computer such as a C server.

An example of a disk subsystem using the level 5 RAID technology will be briefly described with reference to FIG. In FIG. 8, the disk subsystem includes three magnetic disk devices DR1, DR2, DR3 and one spare magnetic disk device DRS. In this figure, illustration of a server computer and a RAID controller is omitted.

[0005] The data managed by the server computer and recorded on the magnetic disk device is divided into predetermined units of blocks, which are distributed and recorded on three magnetic disk devices DR1, DR2 and DR3. At this time, a parity is created as redundant data, and this parity is also recorded in a distributed manner.

[0006] The parity is generated by the following method. The parity P1 is generated by performing an exclusive OR operation on the data Da and the data Db. Similarly, parity P2 is an exclusive OR operation of data Dc and data Dd, parity P3 is an exclusive OR operation of data De and data Df, and parity P4 is data Dg and data Dh.
Is generated by an exclusive OR operation with

In the disk subsystem configured as described above, when a read request for the data Db is made from the server computer, the data Db of the magnetic disk device DR2 is read.
When the data Db cannot be read due to a failure in the recording surface of the medium (magnetic disk) on which the data is recorded or a failure in the entire magnetic disk device DR2, the data read is attempted again and a retry is performed. The RAID controller performs an exclusive OR operation of the data Da and the parity P1 to perform the data Db
Is restored and output as read data to the server computer, and the server computer responds without recognizing that a failure has occurred in the disk subsystem.

When a failure occurs in this way, RAID
The controller is the failed magnetic disk device DR2.
From the RAID configuration and the magnetic disk drive DR1
And DR3.

If a failure occurs in another magnetic disk device in this degenerate state, data cannot be read from the disk subsystem, and the RAID controller notifies the server computer that a failure has occurred in the disk subsystem. Will be.

As described above, it is necessary to restore the RAID to a normal state so that a failure occurs in another magnetic disk device in the degenerate state and no failure occurs in the disk subsystem.

This recovery process is performed by restoring the data of the magnetic disk drive DR2 separated from the spare magnetic disk drive DRS in the degenerate state. Restoring the data of the magnetic disk device DR2 separated from the spare magnetic disk device DRS can be realized by performing an exclusive OR operation on the data of the blocks respectively corresponding to the magnetic disk devices DR1 and DR3. For example, to restore the data De, it can be restored by an exclusive OR operation of the parity P3 and the data Df.

In the recovery processing of a disk subsystem having no spare magnetic disk device, the failed magnetic disk device is replaced with another normal magnetic disk device, and the data is transferred to the replaced magnetic disk device. In the same manner as in the above, restoration is performed by an exclusive OR operation.

[0013]

The disk subsystem using the RAID technology has the following problems. When restoring data of a magnetic disk device disconnected from the RAID configuration due to a failure to a spare or replaced magnetic disk device, the content of the failure of the separated magnetic disk device is a recording surface of a medium (magnetic disk) In order to obtain all data, including data that can be read, by exclusive OR operation, even though data recorded on the medium can be read from a medium where no failure has occurred, It took a lot of time to restore the data.

Also, in such data restoration processing, when data is read out from a magnetic disk device having a RAID configuration and having no failure for exclusive OR operation, a failure occurs. In some cases, data cannot be read. At this time, not only the data restoration process becomes impossible at this time, but also the disk subsystem itself becomes unusable. In this way, it is assumed that the data for which a failure is found during the data restoration process is not very important data that is not normally accessed. The fact that such insignificant data cannot be read out makes it impossible to restore the data of the failed magnetic disk device and renders the disk subsystem itself unusable.

The present invention has been made in order to solve these conventional problems, and has as its object to provide a disk subsystem that can speed up data restoration processing of a failed magnetic disk device.

Further, the present invention is based on the fact that data required for data restoration processing cannot be read from a magnetic disk apparatus in which no failure has occurred during data restoration processing of a failed magnetic disk apparatus. It is therefore an object of the present invention to provide a disk subsystem which prevents a RAID from being unable to be reconfigured and making the disk subsystem unusable.

[0017]

According to the present invention, in a disk subsystem in which a spare magnetic disk device and a plurality of magnetic disk devices constitute a RAID, data of a failed magnetic disk device is stored in the spare magnetic disk device. When restoring to the device, data that can be read from the failed magnetic disk device stores the data in the spare magnetic disk device, and data that cannot be read from the failed magnetic disk device is Data restoration means for restoring by arithmetic operation from data stored in a magnetic disk device in which no failure has occurred among a plurality of magnetic disk devices constituting RAID, and storing the restored data in the spare magnetic disk device Is provided.

According to such a configuration, the data restoration process can be speeded up. Further, according to the present invention, in a disk subsystem in which a spare magnetic disk device and a plurality of magnetic disk devices constitute a RAID, when data of a failed magnetic disk device is restored to the spare magnetic disk device, The data which can be read from the failed magnetic disk device is stored in the spare magnetic disk device, and the data which cannot be read from the failed magnetic disk device is a plurality of magnetic disks constituting the RAID. Data restoration means for restoring the data stored in the magnetic disk device where no failure has occurred in the disk device by operation and storing the restored data in the spare magnetic disk device is further provided. When restoring data, the data used for the calculation is If can not be read from the disk device is characterized in that data of the restoration target is registered in the magnetic disk apparatus of the preliminary and a ECC error.

According to such a configuration, the data restoration process can be speeded up, and at the time of the data restoration process of the failed magnetic disk device, the data necessary for the data restoration process has a failure. RAID cannot be reconfigured due to the inability to read data from a magnetic disk device, and the disk subsystem can be prevented from becoming unusable.

Further, according to the present invention, in a disk subsystem in which a spare magnetic disk device and a plurality of magnetic disk devices constitute a RAID, data of a failed magnetic disk device is restored to the spare magnetic disk device. At this time, of the plurality of magnetic disk devices constituting the RAID, the data stored in the magnetic disk device in which no failure has occurred is restored by calculation, and the restored data is stored in the spare magnetic disk device. When restoring data by the above calculation, if the data used for the calculation cannot be read from the magnetic disk device where the failure has not occurred, it is determined that the data to be restored is an ECC error. A data restoring means for registering in the magnetic disk device.

According to such a configuration, at the time of data restoration processing of a failed magnetic disk device, data necessary for data restoration processing cannot be read from a magnetic disk device having no failure. Can prevent the RAID from being reconfigured and making the disk subsystem unusable.

Further, according to the present invention, a RAID is constituted by a plurality of magnetic disk devices, and when a failure occurs in one magnetic disk device, the failed magnetic disk device is replaced with a new magnetic disk device. In a subsystem, when restoring data of the failed magnetic disk device to the replaced new magnetic disk device, a magnetic disk device in which no failure has occurred among a plurality of magnetic disk devices constituting the RAID. When data is restored by computation from the data stored in the new magnetic disk device and the restored data is stored in the new magnetic disk device, and data is restored by the computation,
If the data used for the calculation cannot be read from the magnetic disk device in which the failure has not occurred, the data restoring means for registering the data to be restored as an ECC error in the replaced new magnetic disk device is provided. It is characterized by having been provided.

According to such a configuration, at the time of data restoration processing of a failed magnetic disk device, data necessary for data restoration processing cannot be read from a magnetic disk device in which no failure has occurred. Can prevent the RAID from being reconfigured and making the disk subsystem unusable.

[0024]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a diagram illustrating a schematic configuration of a system according to the first embodiment of the present invention. FIG. 1 shows a schematic configuration of a computer system including a server computer 100 such as a PC server and a disk subsystem 110.

The server computer 100 has a system bus 10
The system bus 101 is provided with a CPU.
102 is connected. Further, a disk subsystem 110 is connected to the system bus 101 via a SCSI interface (I / F) 111. In addition, SC
SI stands for Small Computer System
This is an abbreviation for Interface.

The disk subsystem 110 includes a CPU 1
12, ROM 113, RAM 114, SCSI interface (I / F) 111 and SCSI interface (I / F).
/ F) 116 is connected to a RAID controller connected via a bus 115 and the SCSI
It is composed of four magnetic disk devices HD1, HD2, HD3 and HDS connected to an I interface (I / F) 116.

The CPU 112 executes the disk subsystem 1 based on the firmware stored in the ROM 113.
10 is controlled in its entirety. The RAM 114 is provided with a conversion table for converting address information when a data read or write command is issued from the server computer 100 to the disk subsystem 110 into address information recorded in an actual magnetic disk device. I have. Also,
The RAM 114 stores various data associated with the operation of the CPU 112. The magnetic disk drive HD1, the magnetic disk drive HD2, and the magnetic disk drive HD3 perform level 5 R.
AID. And the magnetic disk drive HD
S is provided as a spare magnetic disk device.

FIG. 2 shows the configuration of the arrangement of data logically recorded in the disk subsystem 110 as seen from the server computer 100 and the data recorded in the individual magnetic disk devices actually constituting the RAID. The relationship is illustrated.

Data 200 logically recorded in the disk subsystem 110 as viewed from the server computer 100
Is data D divided into six blocks of a predetermined size.
1 to D6. These six data D1 to D
6, three magnetic disk devices HD1 and H
D2 and HD3. Magnetic disk drive HD
1 includes parity data P1, data D3, and data D5.
Is recorded. Similarly, the magnetic disk drive HD2
, Data D2, parity data P2, and data D6 are recorded. Data D2, data D4, and parity data P3 are recorded on the magnetic disk device HD3. Since the magnetic disk device HDS is a spare magnetic disk device, nothing is recorded at this time. The parity P1 is data obtained by an exclusive OR operation of the data D1 and the data D2. Similarly, the parity P2 is data obtained by an exclusive OR operation of the data D3 and the data D4. The parity P3 is data obtained by an exclusive OR operation of the data D5 and the data D6.

The operation of restoring data in the first embodiment of the present invention will be described below with reference to the flowcharts shown in FIGS. 3, 4, and 5. It should be noted that, even during the data restoration operation, instructions for writing and reading data to and from the disk subsystem 110 from the server computer 100 are received in parallel, and the processing is performed. Processing when a data write or read command is received from the server computer 100 during the restoration operation will be described in the following operation description.

Hereinafter, in the disk subsystem 110 shown in FIG.
The data D2 cannot be read from the magnetic disk device HD3 due to a media failure, and the CPU 112 of the RAID control device separates the magnetic disk device HD3 from the RAID configuration to a degenerate state and stores it in the spare magnetic disk device HDS in the magnetic disk device HD3. An operation based on the control of the CPU 112 when restoring the data that has been performed will be described.

First, when restoring data stored in the magnetic disk device HD3 to the spare magnetic disk device HDS, the CPU 112 creates a data change / restore map 300 as shown in FIG. 6 on the RAM 114 (step S1). ). This data change / restore map 300
It is composed of three rows of data 301, status bits 302, and restoration bits 303, and is provided for managing data change and restoration states during the restoration operation of the magnetic disk device HD3 for restoring data. Therefore, D2, D4, and P3 are registered as the data 301.
At the initial stage, the status bit 302 and the restoration bit 3
03 is set to “0”. The restoration bit 303 indicates a state in which the data restoration processing described below is completed for each data (block). This bit is "1"
Means that the corresponding data has been restored to the spare magnetic disk device HDS. The status bit 302 is set when data of a corresponding block is changed by a write command from the server computer 100 during this restoration operation.

Next, the variable B is set to "1" (step S2). This variable B indicates the block number of the data in the magnetic disk device HD3 to be restored. Next, the variable SET, which is the flag information, is set to "1" (step S3). When this variable SET is set to “1”, the CPU 112
This indicates that the data of the block unit D3 is being restored to the spare magnetic disk device HDS. Next, the data of the B-th block is restored (step S4). That is, the data (data D1) of the first block (the block in which the data D1 is stored) is restored.

The details of the restoration process will be described with reference to the flowchart shown in FIG. First, it is checked whether the status bit of the corresponding first block is set to “1” with reference to the data change / restore map 300 (step S50). As a result of this check, if “1” is set in the status bit, step S52
Proceed to. At this time, since the status bit is set to “0”, the process proceeds to step S51.

Next, it is checked whether the data D2 of the first block can be read from the failed magnetic disk device HD3 (step S51). If it can be read, the data D2 is read from the magnetic disk device HD3, copied to the corresponding block of the spare magnetic disk device HDS, and restored (step S53). Then, the restoration operation ends. If the data cannot be read, the process proceeds to step S52.

Next, it is checked whether the data D2 of the first block can be restored by an exclusive OR operation. In other words, it is checked whether the parity P1 and the data D1 as data for performing the exclusive OR operation can be read from the magnetic disk devices HD1 and HD2, respectively (step S52). If it can be read, the process proceeds to step S54, where the parity P1 and data D1
And the data D2 is restored and copied to the corresponding block of the spare magnetic disk device HDS (step S54). Then, the restoration operation ends. If the data cannot be read, the process proceeds to step S55.

In step S55, since the data of the block currently being restored cannot be read from the failed magnetic disk device HD3 and cannot be restored by exclusive OR operation, the spare magnetic disk ECC (Error C) is added to the corresponding block of the device HDS.
[Heating and Correcting] An error has occurred and the spare magnetic disk drive HDS
A flag indicating that an ECC error has occurred is registered in a block corresponding to an error state map (not shown) set in a specific area of the medium. The data restoration process is performed by such a series of operations, and the data restoration process in step S4 in FIG. 3 ends.

Next, the process proceeds to step S5 in FIG. In step S5, since the data D2 has been restored in step S4, the restoration bit corresponding to the data D2 shown in FIG. 6 is set to "1".

Next, in step S6, the variable B is incremented by one. Subsequently, in step S7, the variable SET is set to “0”. Set this variable SET to “0”
Setting this to means that the operation of the CPU 112 restoring the block-based data of the magnetic disk device HD3 to the spare magnetic disk device HDS has been completed.

Subsequently, the process proceeds to step S8, and checks whether the data of the already restored block has been changed by the data write command from the server computer 100. Specifically, in the data change / restoration map 300 shown in FIG. 6, among the data in which the restoration bit 303 is set to “1”, the presence / absence of the data in which the status bit 302 is set to “1” is determined. To check. As a result of this check, if there is data in which both the restoration bit 303 and the status bit 302 are set to “1”, the data is data whose data has already been restored, and the subsequent data from the server computer 100 It is determined that the data has been changed by the data write command, and the process proceeds to step S10. If there is no data in which both the restoration bit 303 and the status bit 302 are set to “1”, the process proceeds to step S9. Hereinafter, the operation when the process proceeds to step S9 will be described, and the operation when the process proceeds to step S10 will be described later.

In step S9, it is checked whether the data of all the blocks has been restored by looking at the value of the variable B. When the data of all the blocks has been restored, all the restoration processing ends. If the data of all the blocks has not been restored, the process returns to step S3 and the above-described operation is repeated.

Next, the CPU 112 issues a command to write or read data from the server computer 100 to the disk subsystem 110 while the above-described steps S3 to S9 are repeated.
In the case where the data is received via the interface 11, the above-described data restoration processing operation is interrupted, the data writing or reading processing is executed, and then the restoration processing operation is restarted.

First, the CPU 112 operates in the server computer 100.
A case in which a data read command is received from will be described. If data is to be read from a normal magnetic disk device constituting the RAID, the data is read.
If a failure occurs and data is read from the magnetic disk device HD3 separated from the RAID configuration,
If the data can be read from the magnetic disk device HD3, the data is read as it is. If the data cannot be read, the data obtained by exclusive OR operation of the data of the corresponding block read from another normal magnetic disk device. Is transferred from the server computer 100 as read data. However, in any of the read cases described above, if data cannot be read due to a media failure, the data is registered as an ECC error as described above, and the server computer 100 notifies the fact.

Next, a case where the CPU 112 receives a data write command from the server computer 100 will be described. First, it is confirmed that the variable SET is “0”. If the variable SET is “1”, “0”
Until it becomes. When it is confirmed that the variable SET is “0”, the following processing is performed. In the case of writing data recorded on a normal magnetic disk device constituting a RAID, the writing process is performed as it is. At this time, it is needless to say that the parity is changed.

Further, in the case of writing data to the magnetic disk device HD3 disconnected from the RAID configuration due to the occurrence of a failure, for example, in the case of writing data D4, the following is performed. First, parity data obtained by performing an exclusive OR operation on the new write data D4 and the data D3 written on the magnetic disk device HD1 is written to the magnetic disk device HD2 as a new parity P2. Then, instead of actually writing the new write data D4, the data change / restore map 300 shown in FIG.
Is set to "1" corresponding to the data D4. The writing process is performed by performing such a process.

Next, in step S8 in FIG. 4, both the restoration bit 303 and the status bit 302 are set to "1", and the data has already been restored, and after that the data from the server computer 100 It is determined that there is data that has been changed by the write command, and step S
The operation when proceeding to step 10 will be described. In the following description, a case where the data change / restoration map 300 is in the state shown in FIG. 4 will be described as an example.

In step S10, a variable SET as flag information is set to "1". Subsequently, step S11
Then, the restoration process of the data D2 for which it is determined in step S8 that both the restoration bit 303 and the status bit 302 are set to "1" is executed again. This restoration processing is performed based on the routine shown in FIG. 7 and has already been described, so that the description is omitted here.

Next, in step S12, the status bit 302 corresponding to the data D2 in the data change / restore map 300 shown in FIG. 6 is set to "0". Subsequently, in step S13, the variable SET is set to "0", and the process returns to step S8. Thus, the restoration bit 303 and the status bit 3
Steps S8 → S10 → S11 → S12 → S13 until there is no more data for which both 02 are set to “1”.
Continue processing.

Next, a second embodiment of the present invention will be described. The difference from the first embodiment is that the operation of the restoration processing shown in FIG. 7 is different, and the other operations are the same as those of the first embodiment.

The operation of the restoration process in the second embodiment will be described with reference to the flowchart shown in FIG. The description of the operation is based on the assumption that the data D2 can be read from the magnetic disk device HD3 due to a media failure in the disk subsystem 110 illustrated in FIG. 1 that configures a RAID of level 5 as in the first embodiment. And the CPU 112 of the RAID control device disconnects the magnetic disk device HD3 from the RAID configuration and enters the degenerate state,
Replace the magnetic disk drive HD with the spare magnetic disk drive HDS.
CPU 11 for restoring data stored in storage 3
The operation based on the control 2 will be described.

First, it is checked whether the data D2 of the first block can be restored by an exclusive OR operation. In other words, it is checked whether the parity P1 and the data D1 as data for performing the exclusive OR operation can be read from the magnetic disk devices HD1 and HD2, respectively (step S60). If it can be read, the process proceeds to step S61, where the parity P1 and the data D
XOR operation with 1 to restore data D2,
The data is copied to the corresponding block of the spare magnetic disk device HDS (step S61). Then, the restoration operation ends.

If reading is not possible, step S
Go to 62. In step S62, it is checked whether the data D2 of the first block can be read from the failed magnetic disk device HD3 (step S62). If it can be read, the data D2 is read from the magnetic disk device HD3, copied to the corresponding block of the spare magnetic disk device HDS, and restored (step S6).
3). Then, the restoration operation ends.

If reading is not possible, step S
Proceed to 64. In step S64, the data of the block currently being restored cannot be restored even by the exclusive OR operation, and the magnetic disk drive
3, the ECC error has occurred in the corresponding block of the spare magnetic disk drive HDS, and the block corresponding to the error state map (not shown) set in the specific area of the medium of the spare magnetic disk drive HDS. A flag indicating that an ECC error has occurred is registered. The data restoration process is performed by such a series of operations, and the data restoration process in step S4 in FIG. 3 ends.

Next, a third embodiment of the present invention will be described. The difference from the first embodiment is that the operation of the restoration process shown in FIG.
If a failure occurs in a magnetic disk device constituting a RAID without having a spare magnetic disk device at 0, the magnetic disk device is to be replaced with a new normal magnetic disk device. Other operations are the same as those of the first embodiment.

The operation of the restoration process in the third embodiment will be described with reference to the flowchart shown in FIG. The description of the operation is based on the assumption that the data D2 can be read from the magnetic disk device HD3 due to a media failure in the disk subsystem 110 illustrated in FIG. 1 that configures a RAID of level 5 as in the first embodiment. The magnetic disk device HD3 is replaced with a new normal magnetic disk device, and the data stored in the magnetic disk device HD3 is restored to the replaced new magnetic disk device (not shown) based on the control of the CPU 112. The operation will be described.

First, it is checked whether the data D2 of the first block can be restored by an exclusive OR operation. In other words, it is checked whether the parity P1 and the data D1 as data for performing the exclusive OR operation can be read from the magnetic disk devices HD1 and HD2, respectively (step S70). If it can be read, the process proceeds to step S71, where the parity P1 and the data D
XOR operation with 1 to restore data D2,
The data is copied to the corresponding block of the replaced new magnetic disk device (step S71). Then, the restoration operation ends.

If reading is not possible, step S
Go to 72. In step S72, since the data of the block currently being restored cannot be restored by the exclusive OR operation, it is determined that an ECC error has occurred in the corresponding block of the replaced new magnetic disk device, and the replaced new magnetic disk A flag indicating that an ECC error has occurred is registered in a block corresponding to an error state map (not shown) set in a specific area of the medium of the apparatus. The data restoration process is performed by such a series of operations, and the data restoration process in step S4 in FIG. 3 ends.

In the above description, since the data of the block being restored cannot be restored, it is determined that an ECC error has occurred, and a flag indicating that an ECC error has occurred in the magnetic disk device that is restoring the data is set. I explained to register. However, if the data that could not be restored is newly written by the server computer 100 thereafter, the registration of the flag may be deleted.

It is presumed that the data in which the flag indicating that the ECC error has occurred is not very important data which is not normally accessed as described in the description of the related art. Therefore, the process of registering that an ECC error has occurred in data that cannot be restored is more difficult than the process of registering that an ECC error has occurred in the data that cannot be restored, as compared with the case where RAID cannot be reconfigured because data cannot be restored and the disk subsystem itself becomes unusable. If you look, it is a better measure.

[0060]

As described above, according to the present invention, the data restoration process can be speeded up, and the data necessary for the data restoration process in the data restoration process of the failed magnetic disk device can be obtained. RAID cannot be read from a magnetic disk device where no failure has occurred.
Can not be reconfigured and the disk subsystem becomes unusable.

[Brief description of the drawings]

FIG. 1 is a diagram showing a schematic configuration of a system according to a first embodiment of the present invention.

FIG. 2 is a view showing the relationship between the configuration of the arrangement of data logically recorded in a disk subsystem as seen from a server computer and the data actually recorded in individual magnetic disk devices constituting a RAID.

FIG. 3 is a flowchart for explaining a data restoration operation according to the first embodiment of the present invention.

FIG. 4 is a flowchart for explaining a data restoration operation according to the first embodiment of the present invention.

FIG. 5 is a flowchart for explaining a data restoration operation according to the first embodiment of the present invention.

FIG. 6 is a diagram showing a data change / restoration map.

FIG. 7 is a flowchart for explaining details of a data restoration process in FIG. 3;

FIG. 8 is a flowchart for explaining details of data restoration processing according to the second embodiment of the present invention.

FIG. 9 is a flowchart for explaining details of data restoration processing according to the third embodiment of the present invention.

FIG. 10 is a diagram illustrating a configuration of an arrangement of data logically recorded in a disk subsystem viewed from a server computer and data actually recorded in individual magnetic disk devices constituting a RAID for explaining a conventional technique; FIG.

[Explanation of symbols]

 100 server computer 110 disk subsystem 111 SCSI interface 112 CPU 113 ROM 114 RAM 115 bus 116 SCSI interface 117 SCSI bus 118 magnetic disk drive HD1 119 ‥‥ Magnetic disk device HD2 120 ‥‥ Magnetic disk device HD3 121 ‥‥ Magnetic disk device HD4

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G06F 12/16 320 G06F 12/16 320L

Claims (8)

[Claims] In a disk subsystem in which a spare magnetic disk device and a plurality of magnetic disk devices constitute a RAID, when restoring data of a failed magnetic disk device to the spare magnetic disk device, Data which can be read from the failed magnetic disk device is stored in the spare magnetic disk device, and data which cannot be read from the failed magnetic disk device is a plurality of magnetic disks constituting the RAID. Data recovery means for recovering data from data stored in a magnetic disk device in which no failure has occurred by operation and storing the recovered data in the spare magnetic disk device. Disk subsystem. 2. The data restoration means according to claim 1, wherein, when restoring the data by operation, if the data used for the operation cannot be read from the magnetic disk device in which the failure has not occurred, the data to be restored is deleted. 2. The disk subsystem according to claim 1, wherein an ECC error is registered in the spare magnetic disk device. 3. A disk subsystem in which a spare magnetic disk device and a plurality of magnetic disk devices constitute a RAID, when data of a failed magnetic disk device is restored to the spare magnetic disk device, Among the plurality of magnetic disk devices constituting the RAID, the data is restored by operation from the data stored in the magnetic disk device where no failure has occurred, and the restored data is stored in the spare magnetic disk device. When restoring data by operation, if the data used for the operation cannot be read from the magnetic disk device where the failure has not occurred, the spare magnetic disk device determines that the data to be restored has an ECC error. A disk subsystem comprising a data restoring means for registering data in a disk subsystem. 4. A disk subsystem in which a RAID is constituted by a plurality of magnetic disk devices, and when a failure occurs in one magnetic disk device, the failed magnetic disk device is replaced with a new magnetic disk device. When restoring the data of the failed magnetic disk device to the replaced new magnetic disk device,
Among the plurality of magnetic disk devices constituting the AID, the data is restored by operation from the data stored in the magnetic disk device where no failure has occurred, and the restored data is stored in the exchanged new magnetic disk device. When restoring data by the above operation, if the data used for the operation cannot be read from the magnetic disk device in which the failure has not occurred, the data to be restored is regarded as an ECC error, and What is claimed is: 1. A disk subsystem, comprising: data restoration means for registering data in a disk device. 5. In a disk subsystem in which a spare magnetic disk device and a plurality of magnetic disk devices constitute a RAID, when restoring data of a failed magnetic disk device to the spare magnetic disk device, Data which can be read from the failed magnetic disk device is stored in the spare magnetic disk device, and data which cannot be read from the failed magnetic disk device is a plurality of magnetic disks constituting the RAID. Data restoration in a disk subsystem, wherein the restoration is performed by operation from data stored in a magnetic disk device in which no failure has occurred, and the restored data is stored in the spare magnetic disk device. Method. 6. When restoring data by the above operation, if data used for the operation cannot be read from a magnetic disk device in which the failure has not occurred, it is determined that the data to be restored is an ECC error. 6. The data restoration method in the disk subsystem according to claim 5, wherein the data is registered in the spare magnetic disk device. 7. In a disk subsystem in which a spare magnetic disk device and a plurality of magnetic disk devices constitute a RAID, when restoring data of a failed magnetic disk device to said spare magnetic disk device, Among the plurality of magnetic disk devices constituting the RAID, the data is restored by operation from the data stored in the magnetic disk device where no failure has occurred, and the restored data is stored in the spare magnetic disk device. When restoring data by operation, if the data used for the operation cannot be read from the magnetic disk device where the failure has not occurred, the spare magnetic disk device determines that the data to be restored has an ECC error. A method for restoring data in a disk subsystem, wherein the data is registered in a disk subsystem. 8. A disk subsystem in which a RAID is constituted by a plurality of magnetic disk devices, and when a failure occurs in one magnetic disk device, the failed magnetic disk device is replaced with a new magnetic disk device. When restoring the data of the failed magnetic disk device to the replaced new magnetic disk device,
Among the plurality of magnetic disk devices constituting the AID, the data is restored by operation from the data stored in the magnetic disk device where no failure has occurred, and the restored data is stored in the exchanged new magnetic disk device. When restoring data by the above operation, if the data used for the operation cannot be read from the magnetic disk device in which the failure has not occurred, the data to be restored is regarded as an ECC error, and A data restoration method in a disk subsystem, which is registered in a disk device.