JP3067558B2 - Extended configuration method of disk array device and disk array device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION This invention is, expanded configuration method and a disk of the disk array device that was scheduled to expansion of disk device
Relates click device, in particular, it is obtained by adding available and without requiring data retraction of existing disk devices.

[0002]

2. Description of the Related Art In recent years, as an external storage device of a computer system, non-volatility of storage, large capacity, high speed of data transfer, and the like are required, and a disk array device has attracted attention as a device satisfying these requirements. FIG. 18 shows a conventional disk array device, in which a disk array controller 12 is provided for a host computer 10 and a disk array 14 in which, for example, five disk devices 1 to 5 are connected in parallel. Here, the disk array has a rank based on the arrangement of a plurality of disk devices constituting the array, and is a port connecting the disk devices 1 to 5. In the case of FIG. 18, the disk array has a five-port one-rank configuration.

[0003] As such a disk array device, RAID1 to RAID5 are conventionally known. Among them, RAID3 and RAID5 are generally put to practical use.
RAID 3 has a fixed disk device for storing parity, and is suitable for accessing large-capacity data such as file data. RAID5 has a distributed disk device for storing parity, and is suitable for transaction processing that requires reading and writing in sector units. The present invention relates to R
A disk array device of AID5 is targeted.

[0004] The capacity of a disk array device is determined by the number of disk devices that are accessed in parallel. Therefore, when it is desired to increase the capacity of the disk array device, it is general to increase the capacity in rank units so as to have a two-rank configuration of the disk arrays 14-1 and 14-2 as shown in FIG. However, since the scale is too large in the case of adding a rank unit, the number of ranks is not changed, and it is necessary to expand the disk unit 6 in units of disk units as shown in FIG.

[0005]

However, when it is desired to expand the disk array device in units of disk devices, the data and parity of the existing disk device are temporarily evacuated to another storage device, and the disk device is expanded. It is necessary to rearrange the data and parity that have been saved later and write them back. For this reason, expansion work cannot be performed while existing data can be accessed, and there has been a problem that the computer system must be stopped with the addition of the disk array device.

Therefore, in a conventional RAID 5 disk array device, data and parity are stored without the need to stop access to existing data as shown in FIG. 21 (see "Nikkei Watcher IBM version 1992.9"). .
No. 14, "IBM 9337 Disk Array Subsystem." That is, data and parity storage addresses (sector addresses) are allocated to the existing disks 1 to 5, and only the data storage addresses are allocated to the additional disk device 6.

For this reason, when the disk device 6 is added, all the storage addresses of the disk device are initialized to zero data, and when the expansion is completed, the address map table including the disk device 6 is changed, so that the existing disk device is replaced. Using the data and parity of the devices 1 to 5 as they are, the access can be shifted to the access including the additional disk device 6.

However, the expansion of the disk array unit of FIG. 21 in units of disk units is based on the premise that an array is composed of five disk units from the beginning, and parity is not stored in an additional disk unit. Eliminating the need to save data and create new parity with the expansion
There is a problem that cannot be applied to expansion with a smaller disk array device of a smaller size.

Further, when the number of additional disk devices increases, the parity is not distributed and stored in the additional disk devices, so that the merit of RAID5 in which different disk devices can be accessed in parallel for the same sector address is reduced. There was also. The present invention has been made in view of such a conventional problem, and it is necessary to add a disk device within the range of a predetermined maximum configuration, to save the data and parity of the existing disk device and to recalculate the parity. It is an object of the present invention to provide an extended configuration method of a disk array device for RAID5, which can be easily performed without any problem.

[0010]

FIG. 1 is a diagram illustrating the principle of the present invention. First, the present invention relates to a disk spare control device 12.
This is an extended configuration method for a disk array device in which N, for example, N = 5 disk devices 1 to 5 are connected in parallel in the maximum configuration.

First, the disk drive shown in FIG.
A virtual address map table 24 shown in FIG. 1 (B) in which addresses for storing data and parity are defined in advance corresponding to the maximum configuration that is used. The initial configuration is configured to be expandable with M disk devices that are less than the maximum configuration number N and are in a range of 3 or more. For example, the minimum configuration is M = 3 disk units 1, 2, and 5.

In the configuration state of M = 3 disk devices 1, 2 and 5, (M-1) = the data and parity storage addresses for the two disk devices 1 and 2 are stored in the virtual address map table 24. The remaining one disk device 5 is defined as a shadow disk device, and the virtual address map table 24 is stored in the shadow disk device.
Creates an address conversion table 26 for allocating a storage address of a parity in which no disk device actually exists.

In response to an input / output request from the host device 10, data is read or written and the parity is updated with reference to the address conversion table 26. First M = 3
When the number of disk devices was increased to (M + 1) = 4, the storage addresses corresponding to the virtual address map table 24 of the added disk device 3 are initialized as shown in the physical address map of FIG. . Next, the parity of the shadow disk device 5 is written back to the additional disk device 3 according to the virtual address map table 24, and the storage address of the shadow disk device 5 after the write-back is initialized.

Further, the address conversion table 26 is defined as (M
+1) = FIG. 1 corresponding to an expanded configuration of four disk devices
The contents are changed to the conversion contents of the physical address map of (D), and data read or data write and parity of (M + 1) disk devices are referred to the changed address conversion table 26 in response to an input / output request from the host device 10. Update.

The virtual address map table 24 is created such that one disk device is used as a spare disk device in an empty state in a maximum configuration using N disk devices. Also, the virtual address map table 24
May be created such that all the disk devices are enabled and the spare disk device is not used in the maximum configuration using the N disk devices.

The virtual address map table 24 stores (N
-1) storing data and parity in one disk device,
The remaining one disk device may be created so that only data is stored. Further, when the number of disk devices reaches the maximum configuration of N units from the expansion, the address translation table 26 is deleted, and the disk device 1
5 to read or write data and update parity.

[0017]

According to the present invention, in a disk array device comprising M disk devices which can be expanded, a virtual address map table in a maximum configuration using N disk devices is provided in a disk array control device (however, provided that a virtual address map table is provided). , 3 ≦ M
≦ N), the data sent from the host device is stored in the physical address of the disk device as it is according to the virtual address map table, and the parity generated from the data is distributed to N disk devices according to RAID5. Is stored. However, since there are only M actual disk devices, parity having no disk device to be stored is based on one of the disk devices where data is to be originally stored as a shadow disk and concentrated on the shadow disk. Store.

Therefore, when a disk device is added, zero data is written to the additional disk device to initialize it, a logical disk device on a virtual address map is allocated, and the parity stored in the additional disk device is transferred to the shadow disk device. Then, zero data is written in the free space created in the shadow disk device after the transfer. Then, the address conversion table is rearranged so as to be compatible with the disk array after the disk devices have been added.

Further, the virtual address map table differs depending on whether or not a spare disk device is provided in the maximum configuration. If the mode with the spare disk device in the maximum configuration is selected, the spare device control unit is started up when N disk devices are prepared, to prepare for occurrence of a failure in the disk array.

[0020]

【Example】

<Table of Contents> (1) Operating environment (2) When spare disk device is not used in maximum configuration (3) When spare disk device is used in maximum configuration (4) Modification when spare disk device is not used in maximum configuration Embodiment (5) Extended Configuration Processing of the Present Invention FIG. 2 is a block diagram showing an embodiment of a disk array device to which the extended configuration method of the present invention is applied.

In FIG. 2, the disk array device of the present invention comprises a disk array control device 12 and a disk array 1.
4 The disk array controller 12 includes a host interface controller 18, an MPU 16, a device interface controller 20, and a data transfer controller 22. The host interface controller 18 of the disk array controller 12 is connected to the host computer 10 via a host interface cable 32.

The device interface control unit 20
Are, for example, five disk devices 1 to 1 provided in the disk array 14 via the device interface cable 34.
5 are connected in parallel. Here, the connection ports of the disk devices 1 to 5 of the disk array 14 are represented by ports P1 to P5. MPU 1 of the disk array controller 12
6, a virtual address map table 24 and an address conversion table 26 are connected. The virtual address map table 24 defines the storage addresses of the disk devices 1 to 5 in sector units in the maximum configuration with five disk arrays 14 as shown in the figure.

For this reason, the MPU 16 includes the disk devices 1 to
In the maximum configuration of 5 units, the requested block is divided into sector blocks on the disk devices 1 to 5 based on an input / output request from the host computer 10 by designating a logical block number, and the sector blocks are allocated. The position is obtained by referring to the virtual address map table 24, and recording or reproduction of data units divided into sector blocks is performed for the corresponding storage addresses in the disk devices 1 to 5.

The address conversion table 26 defines storage sector addresses in an expansion stage until the disk array 14 reaches a maximum configuration of less than five disk arrays. The extended configuration method of the present invention requires at least three disk devices. Therefore, the minimum configuration is such that the disk array 14 has three disk devices, for example, disk devices 1 and 2 and disk device 5. Then, the three disk devices 1, 2 and 5, which are the minimum configuration, can be added one by one in the order of the disk device 3 and the disk device 4.

The address conversion table 26 has a virtual address map table 24 for each of the three disk units in the minimum configuration and the four disk units added by one.
Generated from In general, the extended configuration of the present invention is as follows. When the maximum number of disk devices is N, the number M of real disk devices from three to (N-1) is set to the initial configuration, and M From the initial configuration to the maximum configuration N.

Further, the contents of the virtual address map table 24 provided in the disk array controller 12 differ depending on whether one disk device in the maximum configuration of five devices is used as a spare disk device. That is, when the spare disk device is not used in the five disks having the maximum configuration, the five disk devices 1 to 5 provided in the disk array 14 are used.
, A virtual address map table 24 is created.

On the other hand, when one of the maximum number of five disks is used as a spare disk device, for example, of the five disk devices 1 to 5 provided in the disk array 14, the disk is assigned to the spare disk device. When the device 5 is used, the virtual address map table 24 is created for the remaining four disk devices 1 to 4. (2) When a spare disk device is not used in the maximum configuration FIG. 3 is an explanatory diagram of a virtual address map and a physical address map when a spare disk device is not used in the maximum configuration of five disk arrays 14 in FIG. In this case, as an initial configuration, a case where three disk units 1, 2, and 5 having the minimum configuration are mounted is taken as an example.

FIG. 3A is a virtual address map in which storage addresses are divided into sector units for each of the disk devices 1 to 5 as shown in the figure. In the case of "1-1", the first number indicates a disk device, and the next number indicates a data sector address. Usually, the size of a logical block as an access unit of the host computer is equal to the sector size as an access unit of the disk devices 1 to 5.
For example, if the logical block is 4 KB, the sector size is also 4 KB.

In RAID 5, parity is distributed to different disk devices for different sector addresses. In FIG. 3A, the parity is circulated counterclockwise. That is, the parity P1 is stored in the disk device 5 at the sector address 1, the parity P2 is stored in the disk device 4 at the sector address 2, and the sector address 3 is stored.
To store the parity P3 in the disk device 3, store the parity P4 in the disk device 2 at the sector address 4, store the parity P5 in the disk device 1 at the sector address 5, and return to the disk device 5 again at the sector address 6.

In the virtual address map shown in FIG. 3A, only three disk devices 1, 2, 5 are actually mounted, and the three disk devices 1, 2, 5 are actually mounted.
FIG. 3 for storing data and parity using
The (B) physical address map must be used. The physical address map has the same contents as the virtual address map for the disk devices 1 and 2. In addition, the remaining one disk device 5 is a shadow disk device 5.

The shadow disk device 5 includes a non-existent disk device 3, 4, 5 in the virtual address map.
Parities P1 to P3, P6 to P8 stored in each of
Are stored together. For this reason, even if three disk devices are insufficient for the maximum configuration, all the parities P1 to P8 are actually stored in the parity as in the virtual address map in the maximum configuration. This FIG. 3 (B)
Is obtained from the address conversion table 26 shown in FIG.

FIG. 4 shows initialization processing and parity write back for the additional disk device 3 when one disk device 3 is added in the state where the physical address map of FIG. 3B is mounted. When the disk device 3 is newly added, the disk device 3 is first initialized to zero data. Subsequently, referring to the virtual address map of FIG. 3A, the storage addresses of the parities P3 and P8 allocated to the added disk device 3 are recognized, and stored in the shadow disk device 5 corresponding to the parity storage address. The existing parity P3 and P8 are written back to the added disk device 3.

FIG. 5 shows a state after the parity P3 and P8 have been written back to the added disk device 3. After the write-back is completed, the storage addresses of the sector addresses 3 and 8 of the shadow disk device 5 are zero data. Initialize with Further, the address conversion table 26 in the state shown in FIG. 3B is rewritten in order to enable conversion to the physical address map when the disk device 3 is added. As a result, all the initialization processing by the addition of the disk device 3 is completed, and in the operation state after the addition, the three disk devices 1, 2, 3 and the shadow disk device 5 as shown in FIG. Data and parity are stored.

When it is desired to further add the disk device 4 from the state shown in FIG. 6, the initialization of the additional disk device 4 and the parity P2 and P7 from the shadow disk device 5 are performed in the same manner as shown in FIGS. The write-back and the initialization of the shadow disk device 5 after the write-back are performed, and finally, the address conversion table 26 is rearranged. When the disk device 4 is added, the shadow disk device 5 is eventually used to set the parity P1 and P6 of the disk device 5 in the virtual address map.
, And the other sector addresses are merely 0.

Therefore, the addition of the disk unit 4 is the same as the expansion to the maximum configuration of five units. In this case, the contents of the physical address map are the same as the contents of the virtual address map of FIG. 3A, and therefore, the address translation table 26 is deleted, and the recording of data and parity based on only the virtual address map table 24 is performed. Perform playback. (3) When a spare disk device is used in the maximum configuration FIG.
5 shows a virtual address map when a disk device is used as a spare disk device and a physical address map in an initial configuration when three disk devices are mounted. First, in the virtual address map shown in FIG. 7A, since the disk device 5 is used as a spare disk device, parity and data are not allocated, and the R addresses are assigned to the four disk devices 1-4.
A data storage sector address and a parity storage address are set according to AID5.

FIG. 7B is a physical address map when three disk devices 1, 2 and 5 are mounted. The disk devices 1 and 2 are used as disks for storing data and parity, and the maximum configuration is 5 disks. The disk device 5 which is to be used as a spare disk device is referred to as a shadow disk device 5. Even in this physical address map,
For the disk devices 1 and 2, the same data and parity storage addresses as in the virtual address map are set,
Parity P stored in disk devices 3 and 4 of virtual address map not implemented in shadow disk device 5
1, P2, P5, and P6 are stored together, and zero data is inserted in the address where no parity is stored.

The address conversion table 26 in the initial configuration of the disk devices 1, 2 and 5 incorporates the converted contents of the physical address map shown in FIG. The recording and reproduction of data is performed with reference to the conversion table 26. FIG.
Indicates initialization and parity writeback when the disk device 3 is added. First, all addresses of the added disk device 3 are initialized to zero data. Next, FIG.
Of the virtual address map of the disk device 3 and the parity P2, P6 of the sector address 2 and the sector address 6
The parity addresses P2 and P6 are written back to the additional disk device 3 from the sector addresses 2 and 6 of the shadow disk device.

FIG. 9 shows that the parity P
2 and P6 after being written back, zero data is stored and initialized at sector addresses 2 and 6 of the shadow disk 5 on which the writing back has been completed. And finally, Figure 1
0, a sector address for storing data is allocated to the additional disk device 3, and the address conversion table 26 is rearranged to the contents of the physical address map, thereby completing the extension process. After the extension, the disk array 14 may be accessed using the address conversion table 26 having the contents of the physical address map of FIG. (4) Modified Example When Spare Disk Device is Not Used in Maximum Configuration FIG. 11 shows a virtual address map and a physical address map when the spare disk device is not used in the maximum configuration of five disk devices as in FIG. Yes, as shown in the virtual address map of FIG. 11A, in the maximum configuration, addresses are set so that data and parity are stored in the disk devices 1 to 4 but only data is stored in the disk device 5. It is characterized by the following.

With the setting of such a virtual address map, in the embodiment of FIG. 3, an empty area of zero data is generated in the shadow disk device when the configuration is less than the maximum configuration.
In the embodiment of FIG. 11, the occurrence of free areas in the shadow disk device is eliminated, and the use efficiency of the disk device is increased. FIG. 11B is a physical address map of the mounting state of the three disk devices 1, 2, and 5. For the disk devices 1 and 2, data and parity addresses are set in the same manner as in the virtual address map.
A disk device 5 that stores only data in the virtual address map is assigned to the shadow disk device, and the parity P1, P of the disk device 3, 4 that is not mounted is assigned.
2, P5, and P6, and the storage of the sector addresses 2-3, 1-4, 2-7, and 1-8 that cannot be used in the storage of the parity P3, P4, P7, and P8 in the disk devices 1 and 2, respectively. The address has been set.

Therefore, even if the disk devices 1 and 2 store the parities P3 and P8, the shadow disk device 5
Is present, it is possible to obtain a recording state equivalent to that used for data storage without substantially storing parity. FIG. 12 shows initialization and parity write back when the disk device 3 is added. That is, initialization is performed by putting zero data in all addresses of the added disk device 3. Next, referring to the virtual address map of FIG. 11A, the sector addresses 2 and 6 serving as the parity storage addresses of the disk device 3 are known, and the parity disks P2 and P6 are added from the same sector address by the shadow disk device 5 to the additional disk. Write it back to device 3.

FIG. 13 shows a state after the parity P2, P6 has been written back to the additional disk device 3. When the write-back is completed, the shadow disk device 5 is initialized by putting zero data in the sector addresses 2, 6. . Finally, as shown in FIG. 14, the data storage address 3 is added to the sector addresses 2 and 6 initialized with the zero data of the shadow disk device 5 by referring to the virtual address map of FIG.
-2, 3-6 are set. Then, the address conversion table 26 of FIG. 2 is rearranged so as to have the conversion content of the physical address map of FIG. (5) Extended Configuration Processing of the Present Invention The flowchart of FIG. 15 is a setup processing for creating the virtual address map table 24 and the address conversion table 26 of FIG. First, step S1
Then, the mode of the parity stripe for allocating the parity to a plurality of disk devices is set. This setting mode includes mode 1 to mode 3. Mode 1 is a use mode of the spare disk device (spare disk device) of FIG. 7, mode 2 is a spare disk device non-use mode A of FIG. 3, and mode 3 is a spare disk device in the modified example of FIG. It becomes the non-use mode B.

Subsequently, the process proceeds to step S2, where step S1
The virtual address map table 24 is created in accordance with the contents of the mode setting set in the above. In other words, the virtual address map shown in FIG. 7A is created for setting mode 1, the virtual address map shown in FIG. 3A is created for mode 2, and the virtual address map shown in FIG. A virtual address map of (A) is created.

Next, in step S3, it is checked whether the number of mounted disk devices is less than the maximum configuration N = 5.
If the number is less than five, the process proceeds to step S4 to set the shadow disk device. That is, in the spare disk use mode of mode 1, the spare disk is used as a shadow disk device. In the spare disk non-use mode of mode 2 or mode 3, a specific one is set as the shadow disk device.

Then, the process proceeds to a step S5, at which the address conversion table 26 is created so as to have the contents of the conversion of the physical address map determined by the mounting disk device at that time.
FIG. 16 shows data write processing in response to a write request from the host computer 10 after the setup processing of FIG. 15 has been completed. First, in step S1, when write data is received from the host computer, in step S2, the virtual address map table 24 is referred to, and a sector address for rewriting data is searched for as a parity address of a disk device to which parity storage allocation is performed. I do.

Next, in step S3, the address is converted from the address conversion table 26 into a parity address on the physical address map, and in step S4, the parity is read from the corresponding disk device. Subsequently, in step S5, a new parity is calculated based on the new data and the old parity. In step S6, data is first written to the disk device, and then in step S7, the calculated new parity is written to the disk device. Write.

On the other hand, in the data reading process, reading may be performed by obtaining a sector address on the physical address map with reference to the address conversion table 26. If an uncorrectable error occurs in the read data from the corresponding sector address, the data and parity of the same sector address of another disk device are read, and the read data having an error is restored. This will be transferred to the host computer 10.

FIG. 17 shows an expansion process in the case where a disk device is added while the disk device less than the maximum configuration is mounted. First, in step S1, if one additional disk device is added, the additional disk device is initialized in step S2. That is, all zero data is written to the sector address of the additional disk device. Subsequently, in step S3, it is checked whether the number of mounted disk devices has reached the maximum configuration N = 5.

If the number is less than 5, the address conversion table is rearranged in step S4, the parity is written back from the shadow disk device to the additional disk device in step S5, and the write back is completed in step S6. Zero data is stored in the free space of the shadow disk device.
Here, in the modified embodiment in which the spare disk device of FIG. 11 is not used, a new data storage sector address is set for the sector address of the shadow disk for which write-back has been completed. The actual stored data.

On the other hand, if the number of mounted disk devices reaches N = 5 in the maximum configuration in step S3, the process proceeds to step S7, where the address translation table is deleted, and in step S8, the use mode of the spare disk device is set. Check whether or not. If the mode does not use a spare disk unit,
Proceeding to step S5, the write-back of parity from the shadow disk device to the additional disk device and the storage of zero data after the write-back are performed in the same manner as before.

On the other hand, if the operation mode is the spare disk device use mode, the spare device control unit 30 is started in step S9. In the operation state after the start of the spare device control unit 30, if a failure occurs in any of the four disk devices 1 to 4 used for data storage, the spare device control unit 30 causes the failure. The data of the failed disk device from the data and parity of the remaining three disk devices excluding the failed disk device,
After the restoration is completed, a spare disk unit 5 and three normal disk units constitute a disk array to respond to a request from the host computer 10.

The failed disk device can be used by replacing and repairing the disk device, and then the data of the spare disk device 5 may be transferred, or the disk device replaced with a new disk device is logically set as the spare disk device. May be.
In addition, as a disk device constituting the disk array 14, an appropriate physical device such as an optical disk device, a floppy disk device, and a semiconductor memory device can be used in addition to the magnetic disk device. In the above embodiment, the maximum configuration N
= 5 is taken as an example, but it is needless to say that the number of appropriate maximum configurations can be set as required.

[0052]

As described above, according to the present invention, the number of disk units having the maximum configuration is arbitrarily determined in advance as N, so that the number of disk units from the minimum configuration of 3 to the maximum configuration of N is increased. The disk units can be expanded as needed. This expansion of the disk units does not require the data and parity of the mounted disk units to be evacuated to another disk unit. Can be transferred to the expanded configuration, and the expansion process can be automatically performed by the program sequence of the setup process after the implementation is completed.
It is possible to simply and efficiently add a disk array device.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is a block diagram showing an embodiment of the present invention.

FIG. 3 is an explanatory diagram of a logical address map, a virtual address map, and a physical address map according to an embodiment when a spare disk device is not used in a maximum configuration.

FIG. 4 is an explanatory diagram of initialization of an additional disk device and writing back of parity when the disk device is added in FIG. 3;

FIG. 5 is an explanatory diagram of initialization of the shadow disk device after parity write back following FIG. 4;

FIG. 6 is an explanatory diagram of a physical address map in operation after completion of addition to FIG. 3;

FIG. 7 is an explanatory diagram of a logical address map, a virtual address map, and a physical address map according to an embodiment when a spare disk device is used in a maximum configuration.

FIG. 8 is an explanatory diagram of initialization of an additional disk device and writing back of parity when the disk device is added in FIG. 7;

FIG. 9 is an explanatory diagram of the initialization of the shadow disk device after parity write back following FIG. 8;

FIG. 10 is an explanatory diagram of a physical address map in operation after completion of addition to FIG. 7;

FIG. 11 is an explanatory diagram of a logical address map, a virtual address map, and a physical address map according to a modified example in a case where a spare disk device is not used in the maximum configuration.

FIG. 12 is an explanatory diagram of initialization of an additional disk device and writing back of parity when the disk device is added in FIG. 11;

FIG. 13 is an explanatory diagram of the initialization of the shadow disk device after parity write back following FIG. 12;

FIG. 14 is an explanatory diagram of a physical address map in operation after completion of addition with respect to FIG. 11;

FIG. 15 is a flowchart of a setup process for creating a virtual address map table and an address translation table according to the present invention.

FIG. 16 is a flowchart of a data writing process according to the present invention.

FIG. 17 is a flowchart of an extension process according to the present invention.

FIG. 18 is a block diagram of a conventional disk array device.

FIG. 19 is an explanatory diagram of expansion in rank units.

FIG. 20 is an explanatory diagram of expansion in units of disk devices.

FIG. 21 is an explanatory diagram of data and parity storage states of a conventional device that does not require saving of existing data and parity.

[Explanation of symbols]

 1 to 6: Disk device 10: Host computer (host device) 12: Disk array controller 14: Disk array 16: MPU 18: Host interface controller 20: Device interface controller 22: Data transfer controller 24: Virtual address map Table 26: Address conversion table 28: Parity operation unit 30: Spare device control unit 32: Host interface cable 34: Device interface cable

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-7-160434 (JP, A) JP-A-5-166287 (JP, A) JP-A-5-224832 (JP, A) (58) Field (Int.Cl. ⁷ , DB name) G06F 3/06

Claims (4)

(57) [Claims] 1. A rank of at most N disk units
In the expanded configuration method for a disk array device which constitutes, in constructing the 1 rank N disk apparatuses, to create a virtual address map table that defines the address for storing the data and parity in advance, before Symbol N three or more if you configure one rank Ri by the disk device <br/> of M stand the range of, (M -1) stand storage address is the virtual address map of the data and parity to the disk device in less than Same as table, remaining 1
One disk device is defined as a shadow disk device, and an address conversion table for allocating a storage address of a parity in the virtual address map table where the disk device does not exist to the shadow disk device is created. A data read or data write and parity update with reference to the address conversion table. 2. The method according to claim 1, wherein said M disk devices are (M +
1) When the number of disks has been increased, the storage address corresponding to the virtual address map table of the added disk device is initialized, and the parity of the shadow disk device is written back to the additional disk device according to the virtual address map table. The storage address of the shadow disk device after the return is initialized, the address conversion table is changed in accordance with the expanded configuration of the (M + 1) disk devices, and the address conversion after the change is performed in response to an input / output request from a higher-level device. A method of expanding a disk array device, comprising: reading data or writing data from (M + 1) disk devices and updating parity by referring to a table. 3. A method according to claim 1, wherein said virtual address map table is set to an empty state when one of the disk devices is in an empty state in a maximum configuration using N disk devices. An extended configuration method of a disk array device, which is created to be used as a device. 4. At most one rank in N disk units
The disk array device comprising:
Data and the paris
Disk device that stores the
Equipped with an address map table indicating addresses, one run
Disk drives with less than N disks and 3 or more disks
When configured, it is stored in (M-1) disk devices.
The storage location of data and parity to be
Storage of parity to be stored at the position indicated in the map table and stored in other disk units
The position is set to the remaining one disk device.
Disk array device.