JPH09269871A - Data re-redundancy making system in disk array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data re-redundancy making system in a disk array device which does not lose important data so far as a fault is not generated at another disk at the same time even when a disk fault is generated. SOLUTION: The system is provided with a computer 2 in which a mirroring driver 12 executing redundancy making management to a disk device 6 and a disk driver 14 controlling the disk device 6, and a disk controller 4 inputting and outputting to the disk device. When a fault is generated at a disk B, e.g. the redundancy making of data is executed again so that a disk F of the lowest priority according to a priority order stored in a mirroring managing information table 12a dissolves a mirroring pair with a disk E and forms a new mirroring pair with a disk A. In addition, data is made redundant not only by the unit of the disk but by the unit of the data area and the unit of the sector generating a partial fault.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to redundancy of a disk array device on a computer, and more particularly to a data re-redundancy system for immediately re-redundant with a replacement medium when a redundant medium fails. .

[0002]

2. Description of the Related Art There is a means for implementing redundancy by software and hardware as a disk redundancy means, and there is a disk array device called RAID (cheap disk duplication array) system as a hardware redundancy means. The disk array device is a device for distributing data to a plurality of small disk devices in parallel to several to several tens of disk devices to record data and access them in parallel. When data is transferred to a plurality of disk devices in parallel in the disk array device, it is possible to transfer data at a high speed that is twice as many as the number of disks as compared with the case of one disk device.

Further, by adding and recording redundant information such as parity data in addition to the data, it is possible to detect and correct a data error caused by a failure of the disk device. Tebit A. Patterson and others at the University of California, Berkeley classify the disk array devices that access a large amount of data at high speed into many disks and realize data redundancy in the event of a disk failure from level 1 to level 5. The evaluated paper is published. From level 1 to level 5, RAID (R
edundant Arrays of InexPe
abbreviated as "native Disks" 1 to 5.

RAID 1 is a mirror disk device for writing the same data with two disk devices as a set, and the utilization efficiency of the disk device is low, but it has redundancy.
It is widely used because it can be implemented with simple control.

In RAID 2, data is striped (divided) in units of bits or bytes and read in parallel to respective disk devices. The striped (divided) data is physically recorded in the same sector in all disk devices.

[0006] RAID3 is effective when a large amount of data is continuously handled, but in the case of transaction processing in which a small amount of data is randomly accessed, the high speed of data transfer cannot be utilized and the efficiency is reduced. To do.

In RAID 4, one data is striped in sector units and written in the same disk device. Parity is stored in a definitely determined disk device.
In data writing, since the data and parity before writing are read and the new parity is calculated and written, a total of four accesses are required for one writing. Also,
Since a disk device access for parity always occurs at the time of writing, writing to a plurality of disk devices cannot be executed simultaneously.

RAID 5 enables parallel read / write by not fixing the disk device for parity. That is, the disk device in which the parity is placed is different for each sector. If the parity disks do not overlap, sector data can be written in parallel to different disk devices.

Further, as a disk redundancy means by software, there is a mirroring driver redundancy means. By using the mirroring driver, it is possible to use a generally available disk control device. Further, it is possible to implement redundancy without changing a program called a driver created for controlling the disk control device.

Here, an example of a conventional redundancy system using a mirroring driver will be described. Normally, application software accesses a disk device from an operating system (OS) that integrally manages computers. Since the disk drive control method differs for each disk control device, the disk access method creates application software by absorbing the differences in the disk control device control methods with software called a disk driver and providing a unified interface. Lighten the burden. By using another virtual disk driver between these disk drivers and the operating system, and by performing redundancy within this disk driver, the application software can perform redundancy without being aware of the disk. Can be done. This disk driver is hereinafter called a mirroring driver.

FIG. 12 is a schematic configuration diagram of a conventional computer system for implementing redundancy by using a mirroring driver. The application software 10 in the computer 2 inputs / outputs data to / from the input / output interface provided by the operating system 8. The interface of the mirroring driver 3 is connected to this input / output interface. In the mirroring driver 3, the disk device 6 and the mirroring pairs 68 and 69 are collectively managed. In this conventional example, an input / output interface is created in the operating system 8 corresponding to each mirroring pair. Then, when an access request is made to the input / output interface of the mirroring pair 68, the mirroring driver 3 requests the disk controller 4 for input / output through the disk driver 14. The disk controller 4 inputs and outputs to and from the disks A and B. It should be noted that each component such as the application software 10 on the computer 2 is created by software and is normally installed in the computer 2.
It operates on the above and realizes each function.

[0012]

However, in the conventional redundancy system, when one of the two disks forming a mirroring pair fails during mirroring on the mirroring driver. , The reliability is maintained by using only the failed disk and the normal disk that is being mirrored and operating in a single write, but until the failure is repaired, the remaining normal 1 When the double-writable disk also fails, there is a problem that the data in the disk device is eventually lost.

Also in the RAID system, data reliability is improved between the data disk and the parity disk. However, when a total failure or a partial failure occurs in the data disk or the parity disk, it is the same as the mirroring driver. There was a problem that data was lost when the remaining disks failed because the redundancy became low.

Further, in the RAID system, since the contents of the disk data are not checked and the data is made redundant, there is a possibility that the most important data for the application software may be lost.

The present invention has been made in order to solve the above problems, and an object thereof is not to lose important data even if a disk failure occurs unless another disk also fails at the same time. Another object of the present invention is to provide a data re-redundancy system in a disk array device.

[0016]

In order to achieve the above object, the present invention provides a disk device including a plurality of disks,
Disk control means for performing input / output to / from the disk device, mirroring control means for performing redundancy management such as a mirroring pair for the disk device via the disk control means, and redundancy for storing priority set in the redundancy management unit Characterized in that the mirroring control means performs redundancy of the data contained in the disk to be separated due to a failure or the like according to the priority order.

Further, the redundancy management information storage means stores the priority order set for each disk, and the mirroring control means stores the data contained in the disk when a failure occurs in any of the disks. It is characterized in that the redundancy is performed for each disk according to the set priority order without being limited to the existing mirroring pair.

Further, the mirroring control means performs re-redundancy so as to restore the original mirroring pair when a disk in which a failure has occurred becomes usable.

Further, the redundancy management information storage means stores the priority order set for each user use area, and the mirroring control means is included in a disk when a failure or the like occurs in one of the disks. Data redundancy is performed for each data area according to the set priority without being limited to the existing mirroring pair.

The mirroring control means is characterized in that, when a partial failure occurs in any of the disks, only the data in the data area including the partial failure point is made redundant.

Further, the mirroring control means is characterized by performing re-redundancy so as to restore the original mirroring pair when a disk in which a failure has occurred becomes usable.

Further, according to the present invention, a disk device including a plurality of disks, disk control means for performing input / output to the disk device, and redundancy management for the disk device via the disk control means are performed in block units within the disk. Mirroring control means for carrying out, and redundancy management information storage means for storing redundancy management information set for each block, the mirroring control means, when a failure or the like occurs in any of the disks, It is characterized in that the data contained in the disk is made redundant in block units.

Further, the mirroring control means is characterized in that, when a partial failure occurs in any of the disks, only the data in the block including the partial failure point is made redundant.

Further, the mirroring control means determines a disk having a redundant configuration according to a use environment.

Further, the mirroring control means is characterized by performing re-redundancy so as to restore the original mirroring pair when a disk in which a failure has occurred becomes available.

Further, according to the present invention, a disk device including a plurality of disks, disk control means for performing input / output to the disk device, and mirroring for performing redundancy management such as a mirroring pair for the disk device via the disk control means. A mirroring control information storage means for storing information on a redundantly managed mirroring pair and a spare disk, and the mirroring control means is provided when a failure or the like occurs in any of the disks. It is characterized in that the data contained in the disk is made redundant to the spare disk.

Further, the mirroring control means is characterized in that data redundancy is performed on the spare disk in disk units.

Further, when a partial failure occurs in any of the disks, the mirroring control means copies only the data including the partial failure point to a position corresponding to the partial failure point in the spare disk for redundancy. It is characterized by performing the conversion.

Further, the mirroring control means performs re-redundancy so as to restore the original mirroring pair when a disk in which a failure has occurred becomes usable.

Further, the mirroring control means is characterized in that, when a disk having a failure or the like becomes available, the available disk is used as a spare disk.

Further, the redundancy management information storage means is characterized in that only one of the plurality of disks is used as a spare disk.

Further, the mirroring control means performs re-redundancy so as to restore the original mirroring pair including the disk when the disk in which the failure has occurred is replaced with a usable disk.

Further, the present invention is characterized by having a redundancy processing monitoring means for monitoring the redundancy processing and notifying the user of the processing result.

With the above configuration, even if a disk failure occurs, the important data used by the application software can always be made redundant in disk units or data area units, so that important data is lost. Therefore, the reliability can be improved.

[0035]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below with reference to the drawings. The same components as those in the conventional example are designated by the same reference numerals.

Embodiment 1. FIG. 1 is a schematic configuration diagram showing an embodiment of a computer system adopting a data re-redundancy system according to the present invention, and showing connection states of disk devices before and after a disk failure occurs. The computer system shown in FIG. 1 is composed of a computer 2, a disk control device 4, and a disk device 6 for input / output by the disk control device 4. In the computer 2, the operating system 8 operates as a control program. On operating system 8,
Application software 10 generally called a program operates.

An interface of the mirroring driver 12 is prepared as an input / output interface for the application software 10 to perform input / output to / from the disk device 6. Mirroring driver 12
Is a mirroring control means for performing redundancy management for the disk device 6, and manages the disks A to F included in the disk device 6 as mirroring pairs 61, 62, 63 via the disk control device 4. The mirroring driver 12 performs input / output to / from the disk device 6 via an input / output interface called a disk driver 14 that controls the disk device 6. Therefore,
When an input / output request is made by the application software 10, an input / output is made to the disk device 6 via the disk driver 14 according to the management content of the mirroring driver 12. In FIG. 1, when an input / output request is issued to the mirroring pair 61, the mirroring driver 12 inputs / outputs to / from the disks A and B via the disk driver 14.

Further, the internal data of each of the disks A to F is
It is divided into a management area 6a in which information about mirroring of the mirroring driver 12 is stored and a data area 6b in which actual data is stored. In the management area 6a, information about which disk is a mirroring pair or to what extent the data area 6b has been copied is written.

Further, the computer 2 has a copy monitoring section 16 as a redundancy processing monitoring means for monitoring the redundancy processing such as notification of occurrence of abnormality of input / output to mirroring pair and completion notification of mirroring restoration and notifying the user of the processing result. Mount. The copy monitoring unit 16 is realized as an application interface by software, and its role is to output an error message of the computer 2 to the terminal of the computer 2 when an abnormality occurs in the input / output to the mirroring pair and the end of the mirroring restoration in the mirroring driver 12. (Not shown) and a file. Further, in order to notify the application software 10 of the occurrence of an abnormality, a message is notified via a message transmission interface held by the operating system 8. Further, in order to immediately exchange the disk, a user interface is connected to the copy monitoring unit 16 via a message transmission interface, and the user is notified by sound or light such as a buzzer or an electronic bulletin board.

Further, the mirroring management information table 12a serves as a redundancy management information storage means for storing the priority order set in the redundancy management unit.
It is provided within. The user sets a priority for each mirroring pair. In the first embodiment, priority is set for each of the mirroring pairs 61, 62, 63. In the present embodiment, as shown in FIG.
The priority order is set higher in the order of 61>62> 63. By setting this priority, the data area 6b used by the application software 10 can be given priority.

A feature of this embodiment is that the priority order can be set for the redundant disk device 6. As a result, if you set a higher priority for the disk that stores important data,
When a disk fails, data redundancy is performed again according to the priority order without being limited to the existing mirroring pair, so important data will be reliably redundant and a system with higher reliability will be provided. Can be provided.

As shown in FIG. 1, each component in the computer 5 such as the mirroring driver 12 is represented by software. It goes without saying that various functions are exerted by these software being executed on the CPU of the computer 2, but since the software terms are usually used and are easy to understand, the same applies to this embodiment. explain.

Next, the operation of this embodiment will be described. First, the normal input / output processing will be described.

FIG. 2 shows application software 1
FIG. 9 is a diagram showing an example of the flow of an input / output request when an input / output is requested to the disk device 6 from 0.

When a data input / output request is issued from the application software 10 to the input / output interface of the mirroring pair 61, an input / output request is made to the operating system 8. Since the input / output request to the operating system 8 called the interface of the mirroring pair 61, the input / output request is passed to the mirroring driver 12. The input / output request received by the mirroring driver 12 is interpreted as an input / output request made to the mirroring pair 61 within the mirroring driver 12. Mirroring pair 61
Is composed of the disks A and B, it makes an input / output request to the corresponding disk driver 14.
Upon receiving the input / output request, the disk driver 14 converts the input / output request into an input / output command of the disk controller 6 and executes input / output. Upon receiving the input / output command, the disk controller 4 actually inputs / outputs data to / from the disks A and B.

Here, when a failure occurs on the disk B in the mirroring pair 61 having a high priority, the disk is updated as shown in the mirroring management information table 12a in the mirroring driver 12 shown in FIG. The mirroring pair 63 of E and F having the lowest priority is written as a single write, and the input / output request to the mirroring pair 63 to the disk F is stopped. Then, the data contained in the disk B is made redundant.
Has failed and cannot be actually copied, so the failed disk B and the mirroring pair 61
Therefore, the data of the normal disk A having the above relationship is copied to the disk F. As described above, in this embodiment, in order to prevent the loss of the most important data of the disk A, a new mirroring pair 64 is formed between the disk A and the disk F, and redundancy is performed for each disk. There is. Of course, the disk A may form a new mirroring pair with the disk E which is the disk F and the mirroring pair 63. The mirroring driver 12 copies the entire data area from the disk A to the disk F, manages to which area the data has been copied, and writes the copy information to the management area 6a of the disks A and F. By writing this information, even if the system goes down during the copying process, the copying process can be continued when the computer 2 is restarted. Further, in the present embodiment, since the software called the mirroring driver 12 is used as the mirroring control means to realize redundancy, further input / output requests from other application software 10 are accepted even during the copying process. Can be processed. When the mirroring driver 12 receives an input / output request from another application software 10, if it is a read request, it issues a read request to the disk A, and if it is a write request, it issues a write request to the disks A and F. Performed between copy processes.

After the copying is completed, the mirroring driver 12
Notifies the copy monitoring unit 16 of the end of copying. The copy monitoring unit 16 notifies the application software 10 through the message transmission interface connected to the copy monitoring unit 16. The application software 10 receives the copy end message and outputs the fact that the need for disk replacement has occurred to the system console device and the system error log file.
The user confirms these messages and replaces the disk.

Further, when the disk B fails and the failure further occurs on the disk A in the mirroring pair 64 after the copy is completed, the mirroring pair 63 is already in the disk E.
Since the single writing is performed only for the disk, the remaining mirroring pair 62 is single-written according to the priority order, and the disk C or the disk D is used instead of the failed disk A. In this way, by always making the mirroring pair having a high priority order redundant, the data most important to the application software 10 can be protected from being lost.

The disk exchange method is performed by the disk controller 4
Although different depending on the case, an embodiment will be described here.

In the disk exchange, the disk B in which the error has occurred and the normal disk are exchanged. In the replacement method, the disk B is separated from the disk controller 4. The disk control device 4 has a function of notifying the disk driver 14 that the connection of the disk B is disconnected when the connection is disconnected. Using this function, the mirroring driver 12 detects that the disk B has been disconnected via the disk driver 14. By this detection, the mirroring driver 12 operates the mirroring pair having a low priority as a single write until the disk B is reconnected.

By the way, the mirroring driver 12 is
It can also be applied to the disk control device 4 that does not detect that the disk B has come off. When using the disk control device 4 that does not detect, the user must copy the copy monitoring unit 16
In response, the fact that the disk B has been removed is notified by a command or the like via the message transmission interface held by the operating system 8. By this notification, the mirroring driver 12 recognizes that the disk B has been separated.

Next, after disconnecting the failed disk,
A method for connecting a normal disk will be described.

A function for the mirroring driver 12 to detect that a disk in which a failure has occurred such as the connection of a new normal disk to the disk controller 4 has become available and notify the disk driver 14 is provided. If so, the mirroring driver 12 automatically performs re-redundancy so as to restore the original mirroring pair, thereby performing the recovery process. When the mirroring driver 12 does not have the above function, the user gives a recovery instruction to the copy monitoring unit 16. The restoration instruction is given via the message transmission interface held by the operating system 8. When the copy monitoring unit 16 receives this restoration instruction, the copy monitoring unit 16
The mirroring driver 12 is notified of the restoration instruction, and the mirroring driver 12 carries out restoration processing. The mirroring driver 12 can carry out the recovery process without depending on the disk controller 4 for any normal disk connection method.

Next, the recovery process after the normal disk is connected will be described. Since the mirroring driver 12 stores the initial mirroring pair configuration (disks A and B, disks E and F), when a disk recovery instruction is received from the copy monitoring unit 16, the I / O request to the disk F is stopped. Then, the contents of the disc A are copied to the exchanged disc B. Moreover, the disc contents of the disc E are copied to the disc F. In this way, redundancy can be restarted even for a disk with a low priority. When the copy monitoring unit 16 receives a copy end notification from the mirroring driver 12, the copy monitoring unit 16 notifies the user to that effect. By the series of processes described above, the initial state in this embodiment can be restored.

As described above, according to the present embodiment, when a disk failure or the like occurs, the data redundancy is immediately performed again in accordance with the priority order. Therefore, the disks storing important data must be redundantly configured. Can be
As a result, there is almost no risk of losing data even if the redundant disk fails, and the reliability of the system can be further improved.

Since the purpose of this embodiment is to prevent the loss of important data, it is sufficient to form a mirroring pair with two disks for redundancy. It is possible to multiplex more than triplex.

Embodiment 2. FIG. 3 is the same configuration diagram as that of the first embodiment shown in FIG. 1 except for the mirroring driver 12. The mirroring driver 22 in the present embodiment is characterized in that it does not hold the priority order in the disk unit as in the first embodiment, but holds the priority order in the data area unit used by the user. Therefore, there is a data area in which priority is set in the disk device 6. The disk device 6 shown in FIG.
An example in which the data areas up to is set is shown.

In this embodiment, the user allocates the data areas D1 to D4 in the disk device 6,
The allocated data areas D1 to D4 are used for each application software 10. Data areas D1 to D
By setting 4, the size of the area is secured in advance. Further, the respective data areas D1 to D4 are set so as not to overlap the same disk. At the same time as the area is secured, an input / output interface is created in the operating system 8 as a user interface. When the application software 10 calls the input / output interface of the mirroring driver 22, data input / output is performed in the target data area. Here, input / output interfaces corresponding to the respective data areas D1 to D4 are created. In this example, the size of the area is D4>D2>
The size is secured in the order of D3> D1. Further, the data areas D4 and D2 are secured to be half or more of the disk capacity. The data area D3 is secured by half of the disk. Then, the user sets priorities for the allocated areas. In the second embodiment, the priority order is D1> as shown in the mirroring management information table 22a.
It is set in the order of D4>D2> D3.

Here, it is assumed that a disk error has occurred in the disk B. In the disk B, the data areas D1 and D
Since 4 is set, the mirroring driver 22
Investigates whether another disk has an area of the same size as the data area using the failed disk B. When there is a free area in another disk (in the present embodiment, a free area corresponding to the data area D1 exists in the disk C.), data is transferred from another normal disk A that mirrors the data area. make a copy.
However, if the copy-source disk and the copy-destination disk are the same disk, or if there is no free space for the data area, redundancy is meaningless or copying is not possible. The priority of the data area using B is compared with the priority of the data area of another disk. Then, the size of the low priority data area is calculated, and if the total capacity is larger than the data area in the failed disk, the mirroring of the low priority data area is degenerated. Then, the contents of the normal data area of the disk A having a high priority are copied to the disk which is degenerated and the mirroring is removed. Under the above conditions shown in the present embodiment, the data area D4
If there is a considerable free space on another disk, namely disk C,
Since it does not exist on D, E, and F, the input / output request to the disk F in the data area D3 having the lowest priority is stopped. Then, the contents of the data area D4 on the disk A are copied to the disk device F. As described above, the present embodiment is characterized in that data is copied in units of data areas. As the copy end notification, the copy end notification is sent to the copy monitoring unit 16 as in the first embodiment.

Disc replacement is carried out in the same manner as in the first embodiment. The disk recovery instruction after the disk B is replaced is also executed in the same manner as in the first embodiment. The mirroring driver 22 that has received the restoration instruction stores information about which area is written once and which area is re-mirrored. In the present embodiment, single writing is performed on the data area D3, and the data areas D1 and D
4 is being re-mirrored. Then, with respect to the re-mirrored area, access is stopped and copying is performed on the replaced disk. Further, with respect to the area in which the single writing is performed, the copy is performed again in the area where the re-mirroring is removed. In this embodiment, the contents of the data area D3 of the disc E are copied to the disc F. Regarding the data areas D1 and D4, the input / output request of the data area D1 to the disk C is stopped and the disk F
To the input / output request of the data area D4. Then, the contents of the data areas D1 and D4 on the disk A are copied to the disk B. The copy end notification is given in the first embodiment.
Similarly to the above, a copy end notification is sent to the copy monitoring unit 16.

As described above, according to the present embodiment, the remaining disk capacity and the priority order set for each data area of each disk included in the disk device 6 are carefully controlled so as not to copy to the same disk. When a disk failure or the like occurs, the redundancy is made again, so that the reliability of the system can be further improved.

In the above example, the failed disk B
Although the entire data area included in the disk is copied to another disk to make the data redundant, if the disk B has a partial failure, only the data in the data area including the partial failure point is copied. Alternatively, redundancy may be performed.

Embodiment 3. FIG. 4 is a configuration diagram substantially the same as that of the second embodiment shown in FIG. 3 except for the mirroring driver 32. Unlike the second embodiment, the mirroring driver 32 according to the present embodiment does not hold the priority order in the data area unit, and further subdivides the data area in the disk to make the block unit every several megabytes for redundancy. Is characterized by. Therefore, the disk device 6 has subdivided data areas d1 to d4, and the redundancy management information set for each block is stored in the mirroring management information table 32a as the redundancy management information storage means. Will be.

In the present embodiment, when the user adds data to the disk device 6, the mirroring driver 32 ensures the added data so that it does not overlap the same disk and makes it redundant. In the example shown in the present embodiment, when the area is secured by the application software 10 as the initial setting, the data areas d1 to d4 are secured in blocks. The respective data areas d1 to d4 are secured while being mirrored. The method for determining the area reservation is determined as a usage environment, for example, a disk load, that is, a disk having a small number of disk accesses after the disk is connected to the computer 2. If the disk load is almost the same, judge by the free disk space. Therefore, in this embodiment, it can be said that the priority order is determined by the usage environment such as the disk load and the disk capacity. The disk load and disk capacity can be detected when a disk failure occurs or by regular processing. Of course, the priority order may be predetermined for each block.

When a disk error occurs in the disk B, the data areas used in the disk B in which the failure has occurred, that is, the capacities of the data areas d1 and d4 as shown in FIG. Investigate the disk load of a disk with that amount of free space. Then, if there is a free space corresponding to the area of the failed disk B on another disk, copying is performed according to the disk capacity and the disk load. In the fourth embodiment, the overnight area d1
On the disk C, and the data area d4 on the disk device D. In this way, a disk having a redundant configuration is determined according to the usage environment. If there is no free space on any of the disks, the copy monitoring unit 16
To notify the user that the disk device 6 cannot be made redundant.

Disc replacement is performed in the same manner as in the first embodiment. The disk recovery instruction after the disk replacement is also executed in the same manner as in the first embodiment. Disk B
The mirroring driver 32, which has exchanged and received the restoration instruction, stores which area has been relocated and copied. In this embodiment, the fact that the data area d1 is rearranged in the disk C and the data area d4 is rearranged in the disk D is stored. Then, with respect to the relocated area, access is stopped, and copying is performed on the replaced new disk B. As the copy end notification, the copy end notification is sent to the copy monitoring unit 16 as in the first embodiment.

In the present embodiment, the operation when a partial disk failure occurs will be further described with reference to FIG. FIG. 5 is a diagram showing only the main part of FIG. 4, but the configuration of the disk device is slightly different, and in the example shown in FIG. 5, the areas 1 (disk1, disk2) and the area 2 (disk) are shown.
2, disk3), area 3 (disk2, disk3)
Area 4 (disk1, disk2) is secured on each disk. When the type of failure detected at the time of writing data is a partial failure such as a sector failure, only the data in the block including the partial failure location is copied to the free area of another disk. For example, when a failure occurs in the area 3, the disk 3 having a large number of free areas is targeted for copying at first, but the area 3 already exists.
Investigate isk1. Since there is no area 3 in disk1, the contents of area 3 are copied. The mirroring driver 32 notifies the copy monitoring unit 16 when the copying is completed. The copy monitoring unit 16 notifies the application software through the message transfer interface connected to the copy monitoring unit 16 to determine whether to replace the disk. When there is no free area on another disk, the copy monitoring unit 16 is notified that there is no free area.

The disk replacement when a partial disk failure occurs is carried out in the same manner as in the first embodiment. After the disk is exchanged, the disk recovery instruction is also executed as in the first embodiment. The mirroring driver 32 that has exchanged the disk B and received the recovery instruction
It is checked whether 1 and disk3 are redundant with disk2. Then, the areas are copied in order from the disk 1 to the areas that are not made redundant. As an example, in the restoration method of FIG. 5, areas (1) to (4) are sequentially copied. As the copy end notification, the copy end notification is sent to the copy monitoring unit 16 as in the first embodiment.

As described above, according to the present embodiment, even if a failure occurs in the disk device 6, data redundancy can be performed in block units, so that the reliability of the system is further improved. You can

Fourth Embodiment FIG. 6 shows a fourth computer system adopting the data re-redundancy system according to the present invention.
FIG. 3 is a schematic configuration diagram showing the embodiment of the present invention, in which disks G and H are further connected to the configuration of the first embodiment. The mirroring driver 42 as the mirroring control means in this embodiment has a function of setting a disk as a spare disk when connected, and in this embodiment, the mirroring management information table 4
As shown in 2a, the disks G and H are set as spare disks.

In this configuration, the mirroring driver 4
The normal input / output process for 2 is performed in the same manner as in FIG.
Then, during this normal input / output processing, if the input / output of data to / from the disk B fails, the disk controller 4 notifies the disk driver 14 of an input / output error. Upon receiving the report of the input / output error, the disk driver 14 notifies the mirroring driver 42. In response to this error, the mirroring driver 42 determines that the disk of the disk B has failed.

FIG. 7 is a flow chart showing the internal processing in the mirroring driver 42 when the mirroring driver 42 determines that the disk has failed. When the mirroring driver 42 determines that the disk B has failed (step S1), it determines whether or not a spare disk is registered in the mirroring management information table 42a in the mirroring driver 42 (step S1).
2). If the spare disk does not exist in the mirroring management information table 42a, data cannot be copied. Therefore, the copy monitoring unit 16 is notified and the user is notified that there is no spare disk (step S).
3). If there is a spare disk, the contents of disk A, which is the partner of the mirroring pair of disk B, are copied (step S4).

Next, detailed processing after the occurrence of the failure will be described.

When a failure is detected, the failed disk A
However, the input / output to / from the mirroring pair 61 is stopped. Stop mirroring driver 42
The I / O request for the mirroring pair 61 queued inside is written to the spare disk, and at the same time, the data is copied from the newly mirrored disk A. In this way, data redundancy is performed for each spare disk on a disk-by-disk basis. Mirroring driver 42
Copies the data area from the disk A to the disk G and manages to which area the data has been copied, and writes the copy information to the management area 6a of the disks A and G. By writing this information, even if the system goes down during the copying process, the copying process can be continued when the computer 2 is restarted. Further, in the present embodiment, since software called the mirroring driver 42 is used as the mirroring control means to realize redundancy, further input / output requests from other application software 10 are accepted even during the copying process. Can be processed. Upon receiving an input / output request from another application software 10, the mirroring driver 42 executes a copy request while making a read request to the disk A if it is a read request and a write request to the disks A and G if it is a write request. In between.

By the way, when the detected failure is a partial failure such as a sector error, and when the application software 10 has not yet used the spare disk, only the data including the partial failure spot is set as the partial failure spot in the spare disk. Redundancy is achieved by copying to the corresponding position. The present embodiment is characterized in that copying is performed in units of sectors, and that only the sector including the defective portion is copied to the position of the sector corresponding to the partial defective portion in the spare disk. However, if another disk has already written to the same sector position on the spare disk, writing is performed on the other spare disk. In this way, the failed disk and the spare disk are made to correspond one-to-one and redundancy is provided. FIG. 8 shows an example thereof. First disk2
When a sector error occurs above, the data is copied to the same sector on the spare disk 1 from the disk 1 that holds normal data. Further, the information copied to the spare disk 1 is written into a table and a management area of the disk 1 and the disk 2 which are separately provided in the mirroring driver 42. As a result, as shown in FIG.
Information that the sector corresponding to “k1: sector1” is copied to the sector corresponding to “pre-disk1: setter1” of the spare disk is recorded first in the mirroring driver internal table.

Next, when a sector error occurs on the disk 1, the spare disk 1 is already used by the disk 1, so the information is copied to the same sector on the spare disk 2 and the information copied to the spare disk 2 is stored in the mirroring driver table and the table. Write to the management areas of disk1 and disk2. This is recorded second in the table in the mirroring driver in FIG. Then, when a sector error occurs on the disk 2, redundancy is performed on the spare disk 1 and writing is performed in the management areas of the disk 1 and the disk 2. This is recorded at the third position in the mirroring driver internal table in FIG. When a sector is copied to the spare disk due to a sector error, the mirroring driver 42 causes the copy monitoring unit 16 to
Notify the copy implementation. The copy monitoring unit 16 notifies the application software through the message transfer interface connected to the copy monitoring unit 16 to determine whether to replace the disk.

In the disk exchange, the disk device 15 in which the error has occurred and the normal disk are exchanged. Exchange method is dependent on the disk control method for a disk controller 4, mirroring driver 42 can cope with any replacement method because it implemented by software. After exchanging the disk, the user gives a recovery instruction to the copy monitoring unit 16. The recovery instruction is the operating system 8
Via the message transmission interface owned by. When the copy monitoring unit 16 receives this recovery instruction, the copy monitoring unit 16 notifies the mirroring driver 42 of the recovery instruction. Mirroring driver 42
Is the initial mirroring pair information (disk device 14,
15) is stored, the contents of the spare disk, that is, the new disk in which the mirroring management area and the data area have been exchanged are copied. The copy method is the same as the disk failure. The end of copying can be notified to the user by notifying the copy monitoring unit 16 of the end of copying.

When the disk is replaced at the time of partial failure, the data is copied from the normal disk which is being mirrored. This is shown in FIG. In this figure, disk1 and disk2 are performing a mirroring pair, and disk2 is an example when disk2 is exchanged. If a recovery instruction is issued after disk replacement, di
Data including the management area is copied from the sk1 to the exchanged disk2. However, as mentioned above,
Since a partial failure has occurred also in the disk1, the disk2 cannot be recovered as it is, but in the present embodiment, the sector corresponding to "disk2: sector2" of the disk2 is reserved by referring to the mirroring driver internal table. Disc "pre-disk2: se
It can be seen that the data has been copied to the sector corresponding to "cter2". Therefore, the data of this sector is copied to the disk2, and the spare disk 1 which is no longer used for redundancy and is no longer needed is released.

As described above, according to the present embodiment, the spare disk is prepared, the redundancy is immediately performed when the failure occurs, and the disk redundancy configuration is surely taken. Therefore, the reliability can be improved.

Embodiment 5. The system configuration diagram in the present embodiment is the same as the diagram in the fourth embodiment. Also, the basic operation is almost the same as in the fourth embodiment. However, the present embodiment is characterized in that only one of the plurality of disks mounted in the system is used as a spare disk, thereby making data redundant.

Next, a description will be given of the processing for making the redundancy when a partial failure occurs in any of the disks in this embodiment.

When a failure is detected and the failure is a partial failure such as a sector error, and when the spare disk has not been used yet, only the sector in which the error has occurred is written to the corresponding sector position of the spare disk. After that, even if a sector error occurs in another disk and no spare disk exists, the same spare disk is used in the present embodiment. FIG. 10 is a diagram showing an example thereof. In this FIG.
1 and disk2, disk3 and disk4,
Each forms a mirroring pair.

First, when a sector error occurs on the disk2, the data is written from the disk1 in which the same data is written to the sector position corresponding to the faulty part of the spare disk, and the disk1 and the spare disk are made redundant. Information is written in the management area of the disk1 and the table in the mirroring driver. As a result, as shown in FIG. 10, “disk1: se of disk1 is generated.
The sector corresponding to “cter1” is set to “pre-d” of the spare disk.
The information that the data has been copied to the sector corresponding to “isk1: sector1” is 1 in the mirroring driver internal table.
Recorded in th. Next, when a sector error occurs on the disk1, the disk2 holding the normal data
Copy to the same sector on the spare disk 1. This is recorded second in the table in the mirroring driver in FIG. In FIG. 10, it can be seen that there are records in the mirroring driver internal table until the third and fourth partial failures occur. Then, if another disk has already been used for the same sector of the spare disk, or if there is no spare disk, the mirroring driver 12 notifies the copy monitoring unit 16.

When the disk is replaced after the partial failure such as the sector error occurs, or when the disk is replaced when the partial failure occurs in the sector error, the data is copied from the normal disk which is being mirrored. FIG. 11 shows an example at this time. In this FIG. 11, disk2
Replace the disc. When the recovery instruction is issued after the disk replacement, the disk1 and the disk2 form a mirroring pair.
The data including the management area is copied to k2. However, as described above, since a partial failure has occurred also in the disk1, the recovery of the disk2 cannot be recovered as it is.
Since it can be seen that the sector corresponding to "2: secter2" has been copied to the sector corresponding to "pre-disk2: secter2" of the spare disk, the data in this sector is copied to disk2. Restore the used sectors on the spare disk.

As described above, according to the present embodiment, as long as the sectors do not collide with only one spare disk, the redundancy can be immediately carried out at the time of the failure, and the disk must always be used. Can be made redundant.
Therefore, the reliability of this system can be further improved.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram showing a first embodiment of a computer system adopting a data re-redundancy system according to the present invention.

FIG. 2 is a diagram showing an example of a flow of an input / output request when an input / output is requested to a disk device from application software in the first embodiment.

FIG. 3 is a schematic configuration diagram showing a second embodiment of a computer system adopting a data re-redundancy system according to the present invention.

FIG. 4 is a schematic configuration diagram showing a third embodiment of a computer system adopting a data re-redundancy system according to the present invention.

FIG. 5 is a diagram showing a main part of a computer system in a third embodiment, and is a diagram used for explaining an operation when a partial failure of a disk occurs.

FIG. 6 is a schematic configuration diagram showing a fourth embodiment of a computer system adopting a data re-redundancy system according to the present invention.

FIG. 7 is a flowchart showing processing in the mirroring driver when it is determined that a disk has failed in the fourth embodiment.

FIG. 8 is a diagram showing a main part of a computer system in a fourth embodiment, and is a diagram used for explaining an operation when a partial failure of a disk occurs.

FIG. 9 is a diagram used for explaining a recovery operation after a disk is replaced due to a partial failure in the fourth embodiment.

FIG. 10 is a diagram showing a main part of a computer system in a fifth embodiment, and is a diagram used for explaining an operation when a partial failure of a disk occurs.

FIG. 11 is a diagram used for explaining a recovery operation after a disk is replaced due to a partial failure in the fifth embodiment.

FIG. 12 is a schematic configuration diagram of a conventional computer system that implements redundancy by using a mirroring driver.

[Explanation of symbols]

2 computers, 4 disk control device, 6 disk device, 6a mirroring management area, 6b data area,
8 operating system, 10 application software, 12, 22, 32, 42 mirroring driver, 12a, 22a, 32a, 42a mirroring management information table, 14 disk driver, 1
6 Copy monitoring unit, 61, 62, 63, 64, 65 Mirroring pair.

Claims (18)

[Claims] 1. A disk device including a plurality of disks, a disk control means for performing input / output to and from the disk device, and a mirroring control means for performing redundancy management such as a mirroring pair for the disk device via the disk control means. Redundancy management information storage means for storing the priority order set in the redundancy management unit, wherein the mirroring control means performs redundancy of the data included in the disk to be separated due to a failure or the like according to the priority order. A data re-redundancy method in a characteristic disk array device. 2. The data re-redundancy method in the disk array device according to claim 1, wherein the redundancy management information storage means stores a priority order set for each disk, and the mirroring control means is any one of In the disk array device, when a failure or the like occurs in any of the disks, the data included in the disk is made redundant for each disk according to the set priority without being limited by the existing mirroring pair. Data re-redundancy method. 3. The data re-redundancy method in the disk array device according to claim 2, wherein the mirroring control means re-redundantly restores the original mirroring pair when a disk in which a failure has occurred becomes usable. A data re-redundancy method in a disk array device, which is characterized by performing the redundancy. 4. The data re-redundancy method in the disk array device according to claim 1, wherein the redundancy management information storage means stores a priority order set in a user use area unit, and the mirroring control means comprises: When a failure occurs in any of the disks, the data contained in the disk is made redundant by the data area unit according to the set priority without being restricted by the existing mirroring pair. Data re-redundancy method in array device. 5. The data re-redundancy method in the disk array device according to claim 4, wherein when a partial failure occurs in any of the disks, the data in the data area including the partial failure point is recorded. A data re-redundancy method in a disk array device characterized by performing only redundancy. 6. A data re-redundancy system in a disk array device according to claim 4, wherein said mirroring control means restores the original mirroring pair when a disk in which a failure has occurred becomes usable. A data re-redundancy method in a disk array device characterized by performing re-redundancy. 7. A disk device including a plurality of disks, a disk control means for performing input / output to and from the disk device, and a mirroring control means for performing redundancy management for the disk device in block units within the disk via the disk control means. And a redundancy management information storage unit for storing redundancy management information set for each block, wherein the mirroring control unit is included in a disk when a failure occurs in any of the disks. A data re-redundancy system in a disk array device characterized by performing data redundancy in block units. 8. A data re-redundancy system in a disk array device according to claim 7, wherein said mirroring control means, when a partial failure occurs in any of the disks, only data in a block including the partial failure point is recorded. A data re-redundancy method in a disk array device characterized by performing redundancy. 9. The data re-redundancy method in the disk array device according to claim 7, wherein the mirroring control means determines a disk to be a redundant configuration according to a use environment. Data re-redundancy method. 10. The data re-redundancy system in the disk array device according to claim 7, wherein said mirroring control means restores the original mirroring pair when a disk in which a failure has occurred becomes usable. A data re-redundancy method in a disk array device characterized by performing re-redundancy. 11. A disk device including a plurality of disks, a disk control means for performing input / output to and from the disk device, and a mirroring control means for performing redundancy management such as a mirroring pair for the disk device via the disk control means. Redundant management information storage means for storing information on a redundantly managed mirroring pair and a spare disk, and the mirroring control means is included in a disk when a failure or the like occurs in any of the disks. A data re-redundancy system in a disk array device, wherein data redundancy is performed on the spare disk. 12. The data re-redundancy system in the disk array device according to claim 11, wherein the mirroring control means performs data redundancy for the spare disk in disk units. Re-redundancy method in. 13. The data re-redundancy system in the disk array device according to claim 11, wherein when a partial failure occurs in any of the disks, only the data including the partial failure portion is spared. A data re-redundancy system in a disk array device characterized by performing redundancy by copying to a position corresponding to a partial failure location in a disk. 14. A data re-redundancy system in a disk array device according to claim 11, wherein said mirroring control means restores the original mirroring pair when a disk in which a failure has occurred becomes usable. A data re-redundancy method in a disk array device characterized by performing re-redundancy. 15. The data re-redundancy system in the disk array device according to claim 12, wherein said mirroring control means, when a disk in which a failure has occurred becomes available, A data re-redundancy method in a disk array device characterized by using a spare disk. 16. The data re-redundancy system in the disk array device according to claim 11, wherein the redundancy management information storage means uses only one of the plurality of disks as a spare disk. Data re-redundancy method for disk array device. 17. A data re-redundancy system in a disk array device according to claim 16, wherein said mirroring control means includes an original mirroring including a disk in which a failure has occurred when the disk is replaced with an available disk. A data re-redundancy system in a disk array device characterized by performing re-redundancy so as to return to a pair. 18. A disk re-redundancy system in a disk array device according to claim 1, further comprising a redundancy processing monitoring means for monitoring the redundancy processing and notifying a user of the processing result. Data re-redundancy method in array device.