JP2733189B2 - Disk array device input / output control method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an input / output control method for a disk array device, and more particularly to an operating system for a host computer.
The present invention relates to an input / output control method for a disk array device that can provide a multiplex performance to the disk array device without changing the input / output control portion of the disk array device or making the disk array device multiported.

[0002]

2. Description of the Related Art There has been conventionally known a disk array device which connects a plurality of disk devices connected in parallel to a host computer via a port and reads / writes data from / to the plurality of disk devices in parallel. The configuration of the above disk array device includes a RAID (Redundant Arrays Inexp
Ensive Disks) Level 0 to 5 devices have been proposed (for example, literature: David A. Patterson et al., “A Case for
Redundant Arrays Inexpensive Disks "), RAID level 0 disk array devices that do not have redundancy in the disk devices, and redundancy 1 /
It is said that RAID level 1 is a disk array device having a RAID level of 2, and it is said that the disk array device is not practically used. However, a redundancy 5 in which 5 disk devices are allocated for data and 3 disk devices are allocated for parity. / 8 RAID level 2 disk array device, 4 disk devices are allocated for data, 1 disk device is allocated for parity, 512B per disk and 4KB are provided with 2KB striping sectors. RAI that reads and writes in 2KB units
D level 3 disk array devices have been proposed.

FIGS. 6 and 7 are diagrams showing the configuration of the above-described RAID level 4 and RAID level 5 disk array devices. FIG. 6 shows a RAID level 4 disk array device and FIG. 7 shows a RAID level 5 disk array device. The device is shown. In FIG. 6A and FIG.
0 is the disk array device, 61 is the disk array device 6
0 is a disk device provided in the disk drive shown in FIG.
FIG. 7B shows the arrangement of the sectors on the disk devices # 0 to # 4. The numbers in FIG. 7 indicate the sector numbers, and P indicates the sector storing the parity.

As shown in FIG. 6, a RAID level 4 disk array device has a disk device 61 of # 4 allocated to a parity disk and disk devices 61 of # 0 to # 3 allocated for data. Yes, RA
As shown in FIG. 7, the disk array device of ID level 5 is one in which parity sectors P are distributed and allocated to each of the disks # 0 to # 4, and data is read and written in sector units.

[0005] In the disk array device of RAID level 4 shown in FIG.
Since two disks are allocated to two disks, it is impossible to simultaneously access two disks in a write access in which parity needs to be rewritten. On the other hand, in the disk array device of RAID level 5 shown in FIG. 7, since the sectors P for parity are distributed among the disks # 0 to # 4, in the write access, unless the parity disk is duplicated, 2 It is possible to access two disks simultaneously. For example, sectors with sector number 0 and sector number 6 can be accessed simultaneously because the disk for the parity is a different disk.

By the way, in the above disk array device, the disk array device 60 appears as one device from the host computer to one port, and cannot perform multiplex operation. That is, when only one port is provided in the disk array device, the next access cannot be performed unless the access is completed,
Access to the disk array device cannot be performed simultaneously.

[0007]

As described above, the disk array device appears to the host computer as one device for one port. Therefore, the above-mentioned RAI
In a disk array device of D level 4 or RAID level 5, a plurality of accesses can be issued to one device at the same time, and significant changes such as management of input / output data must be made, or a multi-port must be used. For example, the multiplex performance of the disk array device could not be obtained.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems of the prior art.
To provide an input control method for a disk array device that can derive multiplexing performance from a RAID level 4 or RAID level 5 disk array device without greatly changing the input / output management portion of the disk array device or using multiports. Aim.

[0009]

FIG. 1 is a block diagram showing the principle of the present invention. In the figure, 1 is a host computer, 2 is a RAID level 4 or RAID level 5 disk array device, and the disk array device 2 has n disk devices 3 connected in parallel and n disk devices 3
Is divided into m (m >> n) sectors Sij (i is a disk number, i = 1 to n, j is a sector number of each disk device, j = 1 to m), and each disk device At least one of the three sectors Sik (i = 1 to n) having the same number k is allocated to a parity sector and the remaining sectors are allocated to data sectors, and the parity sector provided in the sector Sik The parity of the data stored in the sector Sik is stored therein, and data is read / written in sector units. 2a is a port provided in the disk array device, and 3 is a disk device.

In order to solve the above-mentioned problem, according to the invention of claim 1 of the present invention, as shown in FIG. 1, in the above-mentioned input / output control method for the RAID level 4 or RAID level 5 disk array device 2, Each of the disk devices 3 is divided into n logical disk units, and logical disk units having the same sector number of each disk device 3 are combined to form n logical disks. When the host computer 1 accesses the host computer 1, the host computer 1 recognizes the logical disk as a different device and reads / writes data, thereby enabling a multiplex operation for the disk array device.

According to a second aspect of the present invention, in the first aspect of the present invention, a parity sector of each logical disk is allocated to a different sector of the disk device 3 for each logical disk. According to a third aspect of the present invention, in the first aspect, the parity sectors of each logical disk are allocated to different disk devices in each logical disk.

[0012]

As shown in FIG. 1, the disk device 3 is divided into logical disk units of the same number, and the logical disk units at the same location of the disk device 3 are combined into one logical disk. Each of # 0 to #n is shown to the upper host computer 1 as one device. And I / O from host computer 1
If there is a request, the disk array device 2 checks which logical disk the I / O request is for, and accesses the disk device 3.

As described above, according to the first aspect of the present invention, each disk device 3 is divided into n logical disk units, and the logical disks having the same sector number of each disk device 3 are divided. Units are combined to form n logical disks, and when the host computer 1 accesses the disk array device 2, the host computer 1 recognizes the logical disks as different devices to read / write data. RAID level 4 or RAI without greatly changing the input / output management portion of the OS of the host computer 1 or making the disk array device 2 multiported.
Multiplexing performance can be derived from the D level 5 disk array device.

According to the second aspect of the present invention, in the first aspect of the present invention, the parity sectors of each logical disk are allocated to different sectors of the disk device 3 for each logical disk. In units of # 0 to #n, RAID level 4 indicates that R
The advantages of AID5 can be realized. According to a third aspect of the present invention, in the first aspect, the parity sector of each logical disk is stored in each logical disk.
Since the disks were allocated to different disk units, the parity disks in each logical disk were distributed and
Access probability can be increased.

[0015]

FIG. 2 is a diagram showing the system configuration of an embodiment of the present invention. In FIG. 2, reference numeral 10 denotes a host computer;
Is a disk array device, and 12 is an array controller for controlling a disk controller. The array controller 12 is provided with a port 12a and has a host computer 1
0 is connected.

Numeral 13 denotes a disk controller for controlling read / write of the disk device, and numeral 14 denotes a disk device. The disk controller 13 has buffers # 0 to # 4 for temporarily holding data at the time of read / write. Is provided. Each disk device 14 is divided into logical disk units of the same number as shown in FIG. In the figure, since there are five disk units 14, each disk unit 14 is divided into five logical disk units.

Then, the logical disk units at the same location of each physical disk device 14 (hereinafter, each disk device 14 is referred to as a physical disk device in order to clarify the difference from the logical disk) are combined into one logical disk. Each of the logical disks # 0 to # 4 is shown to the host computer 10 as one device. That is, from the host computer 10, the logical disk # 0 appears as the first disk device (disk array device), the logical disk # 1 appears as the second disk device (disk array device), and so on.

In order to perform the above control, the array controller 12 is provided with a management table 12b, and the management table 12b stores the logical disk numbers # 0 to # 0.
# 4 and the machine number and the sector number of the physical disk device 14 corresponding thereto, and the physical disk (hereinafter referred to as a parity disk) in which parity is stored in each of the logical disks # 0 to # 4. The machine number and its sector are registered.

When there is an I / O request from the host computer 10, the disk array controller 12 checks which logical disk the I / O request is for, and refers to the management table 12b to access the physical disk to be accessed. And requests the corresponding disk controller 13 to access the disk.

In addition, in each logical disk unit, RAID
As shown in FIG. 4, one of the logical disks is allocated for parity and the other disk is allocated for data. Further, in each of the logical disks # 0 to # 4, the parity disk is allocated to a different disk as shown in FIG. The parity disks must not be the same physical disk device. Therefore, in units of the logical disks # 0 to # 4, R
AID level 4 but overall RAID 5
The advantages of realizing.

As described above, the physical disk device 14 is
Since the logical disks are divided into logical disks 0 to # 4 and the logical disks # 0 to # 4 are viewed as different devices from the host computer 10, the O of the host computer 10
Multiple performance can be obtained without changing the input / output management mechanism of S. For example, even while reading sectors A and B of logical disk # 0 shown in FIG. 7, unless the same physical disk device is accessed, writing of sectors Y and X of logical disk # 3, which is a different logical disk, is possible. is there.

When writing the sectors Y and X, it is necessary to rewrite the parity. Therefore, the parity of the parity disk (physical disk device # 1 in the figure) must be rewritten. Since the disks being read are the physical disk devices # 2 and # 3, the same physical disk device is not accessed, and the reading of the sectors A and B and the writing of the sectors X and Y can be performed simultaneously.

FIG. 3 shows the array controller 1 of the present embodiment.
2 is a flowchart showing an input / output control process in Embodiment 2, and the input / output control process in this embodiment will be described with reference to FIG. When there is an I / O request from the host computer 10, in step S1, it is checked which logical disk the I / O request is for. Further, it determines the parity disk of the logical disk for which the I / O request has been made.

In step S2, it is checked whether the I / O request is a read or a write.
The physical disk device to be accessed by the / O request is determined.
That is, referring to the management table 12b, the physical disk device to read and access the device number and sector of the physical disk is determined, and the parity disk is also accessed at the time of writing. Do.

In step S4, it is checked whether there is a request currently accessing the physical disk device to be accessed obtained in step S3. If there is, the process waits until the request is completed. Next, in step S5, the array controller 12 issues an access request to the disk controller 13, and the disk controller 13 accesses the actual disk in the same procedure as accessing the RAID level 4 disk array device. Do.

In step S6, the array controller 12 performs data transfer with the host in parallel with the disk access by the disk controller 13, or before and after the disk access. FIG. 4 and FIG.
FIG. 6 is a flowchart showing an operation of accessing a disk in the controller 13. The access to the disk will be described with reference to FIGS.

FIG. 4 is a diagram showing processing in write access. In FIG. 4, data of B and C are stored in write access sectors of the physical disk devices # 1 and # 2, respectively. The sectors A and D are stored in sectors corresponding to the sectors where the data B and C are stored in the disk devices # 0 and # 3, and the data A and B are stored in the sectors corresponding to the physical disk device # 4. , C, and D, the physical disk device # 1 and the physical disk device # 2 are accessed at the time of this access.
Shows a case in which data B 'and data C' are written in.

When a write command is issued in step T1 of the figure, the host computer is disconnected in step T2. At step T3, it is determined whether or not data B, C, and P are held in the buffers # 1, # 2, and # 4 of the disk controller by the previous access, and if not, step T4 is performed. And reads data B, C, and P from the write access sectors of the physical disk devices 14 of # 1, # 2, and # 4, and stores the data in the buffers of # 1, # 2, and # 4.
Go to 5.

The disk controllers # 1, # 1
If the data B, C, and P are held in the buffers # 2 and # 4 by the previous access, the process goes to step T5. In step T5, the exclusive OR of the data A and D is obtained from the exclusive OR of the data B and C held in the buffers # 1 and # 2 and the previously obtained parity P held in the buffer # 4. And store it in buffer # 4.

That is, the parity P is the data A to D
Can be obtained from the exclusive OR of the data A and D. By performing the above operation, the exclusive OR of the data A and D can be obtained. In step T6, the host computer 10
Is reconnected, and the data B 'transferred from the host computer 10 in step T7 is stored in the buffer # 1 and the data B' is stored in the buffer # 1.
Write to the physical disk unit of Then, step T8
, Data A and D stored in buffer # 4
And the exclusive OR of the data B 'is obtained and stored in the buffer # 4.

In step T9, the host computer 10
The transferred data C 'is stored in the buffer # 2, and the data C' is written to the physical disk device # 2. In step T10, the exclusive OR of data A, data D, and data B 'stored in buffer # 4 and the exclusive OR of data C' are obtained, and data A, B ', C',
Parity P ′ for D is obtained and stored in buffer # 4,
Write to the # 4 physical disk device.

As described above, buffers # 1, # 2, # 4
When the data B, C, and parity P are stored in the memory, the new parity is obtained using the stored data, so that the new parity can be obtained without reading the data on the disk. FIG. 5 is a diagram showing processing in read access.

In FIG. 3, when a read command is issued in step R1, the host computer 10 is disconnected in step R2, and in step R3, the process waits until data on the disk is ready. When the data on the disk side is ready, the host computer 10 is connected in step R4. In step R5, data is read from the disk, the read data is sent to the buffer, and transferred to the array controller 12. The array controller 12 transfers the data to the host computer 10 in parallel with the reading of the data from the disk.

In the above embodiment, the parity disks of the logical disks # 0 to # 4 are distributed to the physical disk devices of # 0 to # 4.
The disk can be fixed to the physical disk device of # 4. In this case, since it is necessary to write parity to the same physical disk device at the time of writing, it is not possible to perform write access to sectors of different logical disks at the same time. Since it is fixed, the process of determining the parity disk can be simplified.

In the above embodiment, the arrangement of the sectors in each logical disk is RAID level 4, but the arrangement of the sectors in each logical disk is RAID level 5, and the parity disks can be distributed. . In this case, the process of determining the parity disk
Although the number is larger than that of the D level 4, the parity disks in each logical disk are dispersed, so that the probability of simultaneous write access increases.

[0036]

As described above, in the present invention, each physical disk device of a disk array device is divided into logical disk units of the same number, and the logical units of each disk at the same location are collectively combined into one. The following effects can be obtained since the host computer is shown as a disk of the above type. The multiplex performance of the disk array device can be brought out without significantly changing the input / output control portion of the OS of the host computer. Since multiple performance can be obtained without using a multiport disk array device, the load on the disk array controller and its control mechanism can be reduced. Since the multiplex operation of the present invention does not aim at simultaneous access to the same sector, there is no need to provide an additional mechanism such as an exclusive control mechanism for the purpose, and the addition of the disk array controller and its control mechanism is reduced. Can be. By allocating the parity sector of each logical disk to a different physical disk device for each logical disk and distributing the parity, even if each logical disk has a RAID level 4 configuration, the overall
Multiplexing performance equivalent to RAID level 5 can be obtained.

[Brief description of the drawings]

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

FIG. 2 is a diagram illustrating a system configuration according to an embodiment of the present invention.

FIG. 3 is a flowchart showing processing in an array controller.

FIG. 4 is a flowchart showing processing at the time of write access.

FIG. 5 is a flowchart showing processing at the time of read access.

FIG. 6 is a diagram showing a conventional example (1).

FIG. 7 is a diagram showing a conventional example (2).

[Explanation of symbols]

1,10 Host computer 2,11 Disk array device 2a, 12a Port 3,14 Disk device 12 Array controller 13 Disk controller

Claims (3)

(57) [Claims] 1. One port (2a) has n disk devices
(3) are connected in parallel, and the storage areas of the n disk devices (3) are divided into m (m >> n) sectors Sij (where i is the disk number and i = 1 to n, and j is each disk). The sector is divided into j = 1 to m), and at least one of the sectors Sik (i = 1 to n) having the same number k of each disk device (3) is a parity sector. The remaining sectors are allocated to data sectors, the parity of the data stored in the sector Sik is stored in the parity sector provided in the sector Sik, and the data is stored in sector units through the one port (2a). RAID level 4 or RAID level 5 disk array device for reading / writing
In the input / output control method of (2), each disk device (3) is divided into n logical disk units, and the logical disk units having the same sector number of each disk device (3) are put together. When the host computer (1) accesses the disk array device (2), the host computer (1) recognizes the logical disks as different devices and reads / writes data. A method for controlling input / output of a disk array device, wherein a multiplex operation for the disk array device is enabled by performing writing. 2. The input / output control method for a disk array device according to claim 1, wherein a parity sector of each logical disk is allocated to a different sector of a disk device for each logical disk. 3. The input / output control method for a disk array device according to claim 1, wherein the parity sectors of each logical disk are allocated to different disk devices in each logical disk.