JP2001216098A - Disk array device and disk control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a disk array device which can prevent data loss even to the second disk fault occurring during a degenerate operation. SOLUTION: This disk array device 15 is a logical drive including plural disk devices 152 arranged in parallel to each other and can improve its fault resistance by using the redundant data for reproducing the data on a certain device 152 by means of the data on another devices 152. When a degenerate operation is carried out due to the failure of one of devices 152, a host bus adapter(HBA) 151 controlling the device 15 secures one or more idle device 152 among other devices 152 and then reconfigures the new redundant data to reproduce the remaining effective data by using those idle devices 152.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a highly reliable disk array device and a disk control method using the same, and more particularly, to a disk array device and a disk control method capable of preventing data loss even in the event of a further disk failure during degeneration operation About the method.

[0002]

2. Description of the Related Art In recent years, computerization of business has been rapidly progressing in various fields, and a high level of reliability in data processing is required. Under such circumstances, demands for fault tolerance of disks are increasing day by day. RAID technology is known as a method for improving fault tolerance of a disk by having redundant data.

[0003] In particular, RAID5 uses redundant data even when one of a plurality of disk devices constituting a logical drive (array) fails, thereby using the failed disk device. Since the above data can be reproduced, very high fault tolerance is realized.

[0004]

By the way, in the conventional RAID system, even if the disk array has redundant data, when a certain disk device is in a degraded operation due to a failure, the remaining disk array is not operated during the degraded operation. There is a problem that data is lost even if any of the disk devices fails.

For example, in a disk array configuring RAID5, if one disk device fails, it is possible to continue data access by the degenerate operation even if the disk device remains in the same state. In the event of a further failure, data loss will occur immediately.

The present invention has been made in view of such circumstances, and provides a disk array device and a disk control method capable of preventing data loss even in the event of a further disk failure during degeneration operation. The purpose is to do.

[0007]

In order to achieve the above-mentioned object, a disk array device according to the present invention provides a disk array device which shifts to a degenerate operation due to a failure of one of a plurality of disk devices constituting an array. Then, an empty disk device is secured from the remaining disk devices, and new redundant data for reproducing the remaining valid data is further constructed using the secured empty disk device. Yes, for that purpose, a plurality of physically different disk devices are arranged in parallel to form a logically single disk device, and data on one disk device is reproduced using data on another disk device. In a disk array device that has improved fault tolerance by having redundant data, a redundant data is generated when one of the disk devices fails. When the operation shifts to the degenerate operation using data, one or more free disk devices having no valid data are secured from the other disk devices, and the valid data including the redundant data on the other disk devices is secured. It is characterized by comprising a disk control means for constructing new redundant data for reproduction on the reserved empty disk device.

In order to secure an empty disk device from among the remaining disk devices, the disk array device according to the present invention is characterized in that the disk control means (1) compresses valid data on the other disk devices. (2) The valid data is moved so that the free areas of the other disk devices are continuous, or (3) The valid data is moved so that the free areas of the other disk devices are continuous. Later, the effective data after this movement is compressed,
And so on.

In the disk array device according to the present invention, when shifting to the degenerate operation, new redundant data for reproducing remaining valid data (including redundant data for reproducing data on a failed disk device). Therefore, even if one of the disk devices further fails during the degraded operation, all the valid data remaining at the time of the transition to the degraded operation can be guaranteed by using the newly constructed redundant data. Thus, all data during normal operation can be guaranteed.

[0010] The disk array device of the present invention comprises:
The correspondence between the storage positions before and after the compression of the valid data or the correspondence between the storage positions before and after the rearrangement of the valid data is recorded in an address conversion table, and a periodic or predetermined event occurs. Sometimes, based on the address conversion table, it is preferable to rearrange the valid data so that the free areas of the other disk devices are continuous, and furthermore, the virtual data are arranged in virtual address order.

In the disk array device of the present invention, since redundant data can be reconstructed for the disk array during the degenerate operation, even if the disk device fails, data can be accessed without losing data on the array. it can. In addition, by arranging the areas in the order of the virtual addresses, it is possible to improve the performance of accessing a non-failed disk.

[0012]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a configuration diagram of a computer system to which a disk array device according to an embodiment of the present invention is applied.

As shown in FIG. 1, in this computer system, a PCI bus 1 is laid as a system bus. The PCI bus 1 has a CPU 11, a system memory 12, a display controller 13, a keyboard controller 14, and a host bus adapter. (HBA)
151 are connected.

The CPU 11 controls the entire computer system and operates based on the description of an application program including an operating system and utilities stored in the system memory 12.

The system memory 12 is a memory device serving as a main memory of the computer system, and has a CPU
11 stores an application program including an operating system and utilities describing the operations of the eleventh operation.

The display controller 13 controls the output of the user interface of the computer system. The display controller 13 converts the display data in the video RAM 131 drawn by the CPU 11 into a CRT 132 or an LC.
Display control is performed on a display device such as D133.

The keyboard controller 14 controls the input of the user interface of the computer system. The keyboard controller 14 converts the operation of input devices such as the keyboard 141 and the mouse 142 into data and transfers the data to the CPU 11.

The host bus adapter (HBA) 151
In order to configure the disk array device 15 serving as an external storage of the computer system, a plurality of disk devices 152 connected by the SCSI bus 2 are controlled. The disk array device 15 constituted by the host bus adapter (HBA) 151 is a logical drive in which a plurality of disk devices 152 are arranged in parallel.
By having redundant data for reproducing data on one of the disk devices 152 using data on another disk device, the fault tolerance is improved.

In the disk array device 15 of this embodiment, when a further failure of the disk device 152 occurs during the transition to the degenerate operation using the redundant data due to the failure of one of the disk devices 152, However, the feature is that the disk access can be continued by preventing data loss, and this point will be described in detail below.

(First Embodiment) First, a first embodiment of the disk control in the disk array device 15 will be described. FIG. 2 is a diagram showing a configuration of the host bus adapter (HBA) 151 according to the first embodiment.

As shown in FIG. 2, the host bus adapter (HBA) 151 includes an array controller 1511,
It has a SCSI control mechanism 1512, a failure recovery function 1513, and an address conversion table 1514.

The array controller 1511 is the core of the disk array device 15 and not only executes data arrangement control and redundant data generation during normal operation, but also performs data reproduction during degeneration operation and Perform the disk control specific to the present invention as shown.

That is, the array controller 1511
As shown in FIG. 3, when a failure occurs in a disk constituting an array (FIG. 3A), an empty disk is secured by compressing data on other disks (FIG. 3 (A)). b)), and use this blank disk to further construct redundant data relating to data on other disks (FIG. 3 (c)).

Further, the SCSI control mechanism 1512
Data communication with the disk device 152 is performed based on a protocol defined by the SI bus 2. When the failed disk device 152 is replaced with a new disk device 152, the failure recovery mechanism 1513 reconstructs data that the replaced disk device 152 should have during normal operation independently of the array controller 1511. I do. The address conversion table 154 records address information for implementing disk control unique to the present invention during the degenerate operation.

The array controller 1511 compresses data of the active array (N disk devices) in the degenerate state, and places all valid data on the array on N-1 disks. Then, the redundant data for the data on the N-1 disk devices having the valid data is reconstructed on one disk having no valid data. If the data capacity after compression is insufficient for N-1 disks, the redundant data is not reconstructed and the disk is operated in the degenerate mode.

This compression is performed on a fixed data block (sector or the like) basis. After the data block is compressed, the data size is smaller than the original data size. However, as shown in FIG. 4, an area is allocated on the array without aligning the data block size. For this reason, even when the file system accesses one data block, there are cases where data compressed into a plurality of (consecutive) data blocks actually exists on the array.

At the time of compression, the disk controller records the correspondence between the data block address seen from the file system and the actual data block on the array in the address conversion table 1514 described above.

When a disk read is requested from the host side, the array controller 1511 reads out the compressed data while referring to the address conversion table 1514, expands the read data, and expands the expanded data. Hand over to the host side.

On the other hand, when writing to the disk is requested from the host, the array controller 1511
First determines whether or not old compressed data corresponding to the requested write block exists on the disk array while referring to the address conversion table 1514, and if so, discards the old data, The block in which this data was recorded is defined as a free area.

Next, the requested data is compressed and written in a free area on the disk array, and the newly written position is recorded in the address conversion table 1514. FIG.
FIG. 4 is a diagram showing the movement of blocks due to this writing.

As shown in FIG. 6, the array controller 1511 periodically or when there is no remaining area for writing the compressed data on the disk array, as shown in FIG. The data area is reallocated (garbage collected) so as to be continuous.
At this time, as shown in FIG. 7, data area reassignment (garbage collection) may be executed while taking into consideration the order of virtual addresses.

When a write request is issued from the host, if an area for writing compressed data cannot be newly secured on the array despite the reallocation, the host Notifies the other side that the write request has failed.

As a result, during the degenerate operation, as long as a new redundant disk can be secured by compressing data, even if another disk failure occurs, disk access can be continued. .

Next, an operation procedure of the disk array device 15 will be described with reference to FIGS.

When a failure occurs in the disk device 152, the array controller 1511 determines whether or not there are two or more non-failed disk devices 152 (step A1 in FIG. 8). Y in step A1
ES), and constructs the address conversion table 1514 while compressing these valid data (step A in FIG. 8).
2).

Next, the array controller 1511 determines whether a disk device 152 having no valid data has been secured by this compression (step A3 in FIG. 8), and if it has been secured (YES in step A3 in FIG. 8). Then, redundant data relating to data on the other disk devices 152 is formed on the disk device 152 having no valid data (step A4 in FIG. 8).

When a read request occurs during the degenerate operation, the array controller 1511 determines whether or not the data has been compressed (step B in FIG. 9).
1) If not compressed (N in step B1 in FIG. 9)
O), read data from the requested address (FIG. 9)
Step B2).

On the other hand, if the data has already been compressed (YES in step B1 in FIG. 9), the array controller 1511 determines the position of the compressed data with reference to the address conversion table (step B3 in FIG. 9), and The expanded data is read and expanded (step B4 in FIG. 9), and the requested data is transferred to the host from the expanded data (step B5 in FIG. 9).

When a write request occurs during the degenerate operation, the array controller 1511 determines whether the data has been compressed (step C in FIG. 10).
1) If not compressed (N in step C1 in FIG. 10)
O), write data to the requested address (FIG. 10)
Step C2).

On the other hand, if the data has already been compressed (YES in step C1 in FIG. 10), the array controller 1511 compresses the write data (step C3 in FIG. 10).
It is determined whether or not there is a free area for writing the compressed data (step C4 in FIG. 10).

If there is no free area (NO in step C4 in FIG. 10), the array controller 1511
The data area is reallocated (garbage collect) (step C5 in FIG. 10), and it is determined whether a free area is secured (step C6 in FIG. 10). If no free area is secured (step C in FIG. 10)
6: NO), this write request is terminated with an error.

When there is a free area (YES in step C4 in FIG. 10) and when a free area is secured by reallocation (garbage collection) (YES in step C6 in FIG. 10), the array controller 1511
The corresponding old compressed data on the array is discarded with reference to the address conversion table, and the newly compressed data is written in a free area on the array to update the address conversion table (step C7 in FIG. 10).

As described above, this disk array device 1
5 makes it possible to prevent data loss even in the event of further disk failure during the degenerate operation.

(Second Embodiment) Next, a second embodiment of disk control in the disk array device 15 will be described.

FIG. 11 is a diagram showing the configuration of the host bus adapter (HBA) 151 according to the second embodiment.

As shown in FIG. 11, this host bus adapter (HBA) 151 has an effective block table 1515 in addition to the configuration of the host bus adapter (HBA) 151 shown in FIG. 2 in the first embodiment. It is newly provided. This effective block table 1515
Is used to record an area where valid data is stored. In the second embodiment, the address conversion table 1514 is used not to record the correspondence between before and after the compression of valid data, but to record the correspondence between before and after the movement of valid data, which will be described later. Used for

In the second embodiment, as shown in FIG. 12, the array controller 1511 operates when a failure occurs in a disk constituting the array (see FIG. 12).
(A)), an empty disk is secured by moving data on the other disks (FIG. 12 (b)), and redundant data relating to data on the other disks is further constructed using the empty disks. (FIG. 12 (c)).

The array controller 1511 monitors disk writing after the initialization of the disk array, and validates all areas in the array that have been used for a file system (assuming that valid data is stored). Record in the block table 1515.

When the degenerate mode is entered (operating with N disk devices), the data on all the blocks recorded in the effective block table is transferred to (N-1) active disk devices. After the movement, the correspondence between before the movement and after the movement is recorded in the address conversion table 1514 described above. As a result, one or more disk devices 152 having no valid data are secured.

If a disk device having no valid data can be secured, redundant data for data on the remaining N-1 disk devices is reconstructed here. If the capacity of the N-1 disks is insufficient, the redundant data is not reconstructed, and the disk operates in the degenerate mode.

When reading is requested from the host side,
The array controller 1511 specifies the position of data on the array with reference to the address conversion table 1514,
The requested data is read. On the other hand, when writing is requested from the host side, the array controller 151
1 is the address translation table 1514 for writing to a block registered in the address translation table 1514.
For writing to an area specified by the address conversion table 1514 and writing to a block not registered in the address conversion table 1514, an invalid block is newly secured from the areas on the N-1 disk devices, and the address conversion table 1514 is obtained. And write the specified data into this area. If a new area cannot be secured, the host is notified that the write request has failed.

As a result, during the degenerate operation, as long as a new redundant disk can be secured by moving data, even if another disk failure occurs, disk access can be continued. .

Next, an operation procedure of the disk array device 15 will be described with reference to FIGS.

When a failure occurs in the disk device 152, the array controller 1511 determines whether or not there are two or more non-failed disk devices 152 (step D1 in FIG. 13). Step D1
YES), the address conversion table 1514 is constructed while transferring the data registered in the effective block table (step D2 in FIG. 13).

Next, the array controller 1511 determines whether a disk device 152 having no valid data has been secured by this movement (step D3 in FIG. 13), and if it has been secured (YES in step D3 in FIG. 13). Then, redundant data relating to data on the other disk devices 152 is formed on the disk device 152 having no valid data (step D4 in FIG. 13).

When a read request occurs during the degenerate operation, the array controller 1511 determines whether or not the data has been moved (step E in FIG. 14).
1) If not already moved (N in step E1 in FIG. 14)
O), read data from the requested address (FIG. 1)
Step E2).

On the other hand, if the data has been moved (YES in step E1 in FIG. 14), the array controller 1511 determines the position of the compressed data with reference to the address conversion table 1514 (step E3 in FIG. 13). The requested data is read from the destination address and delivered to the host (step E4 in FIG. 14).

When a write request occurs during the degenerate operation, the array controller 1511 determines whether or not the data has been moved (step F in FIG. 15).
1) If not moved (N in step F1 in FIG. 15)
O), write data to the requested address (FIG. 15)
Step F2).

On the other hand, if the data has been moved (YES in step F1 in FIG. 15), the array controller 1511 determines whether or not there is a free area for writing the requested data (step F3 in FIG. 15). .

Here, if there is no free area (NO in step F3 in FIG. 15), this write request ends in error. On the other hand, when there is a free area (YES in step F3 in FIG. 15), the array controller 1511
The corresponding old data on the array is discarded with reference to the address conversion table 1514, and the requested data is written in a free area on the array to update the address conversion table 1514 (step F4 in FIG. 15).

As described above, this disk array device 1
5 makes it possible to prevent data loss even in the event of further disk failure during the degenerate operation.

In the above-described embodiment, an example has been described in which valid data on a disk device other than the failed disk device is compressed or moved to secure a free disk device. If the compression is further performed after the compression, the possibility of securing an empty disk device can be further increased.

[0063]

As described in detail above, the disk array device of the present invention can be used when (1) other operations are performed when the disk array device shifts to the degenerate operation due to the failure of one of the plurality of disk devices constituting the array. (2) move the valid data so that the free areas of the other disk devices are continuous, or (3) move the valid data so that the free areas of the other disk devices are continuous. After the data is moved, the effective data after the movement is compressed, etc., to secure an empty disk device from among the remaining disk devices, and the remaining empty disk device is used to secure the remaining disk device. New redundant data to regenerate valid data
When shifting to the degenerate operation, new redundant data for reproducing remaining valid data (including redundant data for reproducing data on the failed disk device) is further constructed. Even if one of the disk units further fails, it is possible to guarantee all remaining valid data at the time of degraded operation transition using newly constructed redundant data, thereby guaranteeing all data during normal operation. Become.

The correspondence between the storage positions of the effective data before and after the compression or the correspondence between the storage positions of the effective data before and after the rearrangement is recorded in the address conversion table, and is periodically or predetermined. When the event occurs,
Based on the address conversion table, the data on the failed disk device is reproduced by rearranging the valid data so that the free areas of the other disk devices are continuous and further in the order of the virtual address. Overhead is as small as possible.

[Brief description of the drawings]

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

FIG. 2 is a diagram showing a host bus adapter (H) according to the first embodiment;
FIG.

FIG. 3 is a diagram showing a host bus adapter (HB) according to the first embodiment;
FIG. 2A is a diagram for explaining an outline of disk control performed by the array controller of FIG.

FIG. 4 is a diagram showing a host bus adapter (HB) according to the first embodiment;
FIG. 2A is a first diagram for explaining data compression performed by the array controller of FIG.

FIG. 5 is a diagram illustrating a host bus adapter (HB) according to the first embodiment;
FIG. 2A is a second diagram illustrating data compression performed by the array controller of FIG.

FIG. 6 is a diagram showing a host bus adapter (HB) according to the first embodiment;
FIG. 3A is a first diagram for explaining relocation of compressed data performed by the array controller of FIG.

FIG. 7 shows a host bus adapter (HB) according to the first embodiment;
FIG. 2A is a second diagram for explaining relocation of compressed data performed by the array controller in FIG.

FIG. 8 is a diagram showing a host bus adapter (HB) according to the first embodiment;
4A is a first flowchart for explaining the operation of the array controller during the degenerate operation.

FIG. 9 is a diagram showing a host bus adapter (HB) according to the first embodiment;
9A is a second flowchart for explaining the operation of the array controller during the degenerate operation.

FIG. 10 shows a host bus adapter (HB) according to the first embodiment.
3A is a third flowchart for explaining the operation of the array controller during the degenerate operation.

FIG. 11 is a diagram showing a configuration of a host bus adapter (HBA) according to the second embodiment.

FIG. 12 shows a host bus adapter (HB) according to the second embodiment.
FIG. 2A is a diagram for explaining an outline of disk control performed by the array controller of FIG.

FIG. 13 shows a host bus adapter (HB
4A is a first flowchart for explaining the operation of the array controller during the degenerate operation.

FIG. 14 shows a host bus adapter (HB) according to the second embodiment.
9A is a second flowchart for explaining the operation of the array controller during the degenerate operation.

FIG. 15 shows a host bus adapter (HB) according to the second embodiment.
3A is a third flowchart for explaining the operation of the array controller during the degenerate operation.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... PCI bus 2 ... SCSI bus 11 ... CPU 12 ... System memory 13 ... Display controller 14 ... Operation continuation module 15 ... Disk array device 131 ... Video RAM 132 ... CRT 133 ... LCD 141 ... Keyboard 142 ... Mouse 151 ... Host bus adapter (HBA) 152 ... Disk device 1511 ... Array controller 1512 ... SCSI control mechanism 1513 ... Fault recovery mechanism 1514 ... Address conversion table 1515 ... Effective block table

Claims (13)

[Claims] A plurality of physically different disk devices are arranged in parallel to constitute a logically single disk device, and data on a certain disk device is reproduced using data on another disk device. In a disk array device having improved fault tolerance by having redundant data, when one of the disk devices causes a failure and shifts to a degraded operation using the redundant data, the other disk device Vacant disk device in which one or more vacant disk devices having no valid data are reserved from among the vacant disk devices and new redundant data for reproducing valid data including the redundant data on the other disk devices is reserved. A disk array device comprising a disk control means constructed above. 2. The disk array device according to claim 1, wherein said disk control means secures said empty disk device by compressing valid data on said other disk device. 3. The disk array device according to claim 2, wherein said disk control means records a correspondence relationship between storage positions of the valid data before and after the compression in an address conversion table. 4. The relocation means for periodically rescanning the address conversion table and relocating the compressed valid data so that free areas of the other disk devices are continuous. 4. The disk array device according to claim 3, comprising: 5. The disk control means scans the address conversion table when a shortage of free space is detected at the time of a write process, and performs the compression after the compression so that the free space of the other disk device is continuous. 4. A relocation means for relocating valid data.
The disk array device according to the above. 6. The disk array device according to claim 4, wherein said rearrangement means rearranges the compressed valid data so as to be in virtual address order. 7. The disk controller according to claim 1, wherein the disk controller secures the empty disk device by moving the valid data so that the empty regions of the other disk devices are continuous. Disk array device. 8. The disk array device according to claim 7, wherein said disk control means records the correspondence between the storage positions of the valid data before and after the movement in an address conversion table. 9. The disk array device according to claim 8, wherein said disk control means moves said valid data so as to be in virtual address order. 10. The disk control means, after moving the valid data so that the free areas of the other disk devices are continuous, compresses the valid data after the movement to reduce the free disk device. 2. The disk array device according to claim 1, wherein the disk array device is secured. 11. A system in which a plurality of physically different disk devices are arranged in parallel to logically constitute a single disk device, and data on a certain disk device is reproduced using data on another disk device. In the disk control method for a disk array device having improved fault tolerance by having redundant data, when a failure occurs in one of the disk devices and the system shifts to the degraded operation using the redundant data, By compressing the valid data on the other disk devices, one or more free disk devices having no valid data are secured, and a new data for reproducing the valid data including the redundant data on the other disk devices is obtained. A disk control method, wherein redundant data is constructed on the reserved empty disk device. 12. A plurality of physically different disk devices are arranged in parallel to form a logically single disk device, and data on one disk device is reproduced using data on another disk device. In the disk control method for a disk array device having improved fault tolerance by having redundant data, when a failure occurs in one of the disk devices and the system shifts to the degraded operation using the redundant data, By moving valid data so that free areas on other disk devices are continuous, one or more free disk devices having no valid data are secured, and the valid data including the redundant data on the other disk devices is secured. Disk control method for constructing new redundant data for reproducing a disk on the reserved empty disk device 13. A system in which a plurality of physically different disk devices are arranged in parallel to logically constitute a single disk device, and data on one disk device is reproduced using data on another disk device. In a disk control method for a disk array device that has improved fault tolerance by having redundant data, when a failure occurs in any one of the disk devices and the system shifts to a degraded operation using the redundant data, After moving the valid data so that the free areas on the other disk devices are continuous, one or more free disk devices having no valid data are secured by compressing the valid data after the movement, and A new redundant data for reproducing valid data including the redundant data on the disk device is constructed on the secured empty disk device. Disk control method comprising Rukoto.