JPH07134637A - Disk array device - Google Patents

Abstract

(57) [Summary] [Purpose] To provide a disk array device that, when data update in sector units occurs, performs data update in sector units in a multiplexed manner to speed up data update processing. [Configuration] To generate data and check data transferred from the host computer so that the disk array control device 9 can perform multiple data transfer to each of the disk drives D0 to D3 and check data generation processing accompanying data update. Buffer memories 10a to 10d for storing necessary data corresponding to the disk drives are provided, and each disk drive is provided with controllers 12a to 12d for controlling the positioning of the disk drives and data transfer, and high-speed access as a check disk drive D3. The storage means capable of achieving the above is provided, for example, a semiconductor disk having a backup function by a non-volatile mechanism. In addition, T is a transceiver.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array device having a check function of data recorded in a plurality of disk drives and storing the check data in the same disk drive.

[0002]

2. Description of the Related Art Conventionally, as shown in FIG. 3, a disk array device comprises a plurality of disk drives D0 to D3 (here, four as an example) and a shared controller 14. The controller 14 uses the disk drives D0 to D
3 and controls an interface between each disk drive D0 to D3 and the host computer. A normal disk array device uses one of the plurality of disk drives D0 to D3 as a checking disk drive and has an error checking function using parity data.

Here, as shown in FIG. 3, the recording medium of each of the disk drives D0 to D2 for data has logical sector addresses 0 to 11 in addition to physical sector addresses 0 to 3.
To set. When the disk drive D3 is used as a check disk drive, the controller 14
Check data corresponding to each data of the same physical sector written in each disk drive D0 to D2 is generated,
The data is stored in the same physical sector of the check disk drive D3. By the way, the sector unit data read processing of each disk drive D0 to D2 can be performed in multiplex because only the data is read from each disk drive D0 to D2.

However, in the control of the array disk shown in FIG. 3, when the data update of the logical sectors 0 and 7 occurs, the writing to the disk drives D0 and D1 can be performed in a multiplexed manner, but the check data is generated. However, the process of storing in the check disk drive D3 cannot be performed in multiple because it is the same disk drive. As described above, the method of storing the check data in the check data disk drive has a problem that the access is concentrated on the check data disk drive when the sector-based data update frequently occurs.

As a solution to this problem, as shown in FIG. 4, a method is devised in which check data are stored in the disk drives D0 to D3 while being shifted for each physical sector to prevent access from being concentrated on a specific disk. ing. When the data update of the logical sectors 0 and 7 occurs by this method, the writing to the disk drives D0 and D2 is performed in a multiplexed manner, and the check data update is also performed on the disk drive D3.
And D1 are distributed, so multiple processing can be performed.

[0006]

However, in the conventional array disk device shown in FIG. 4 as well, considering a logical sector in which data update can be multiplexed in an area in which the disk drive for storing check data makes one round, the disk drive is considered. When the data update of the logical sector 0 of D0 occurs, only the sector 4 or 7 can perform the data update in multiplex except the same physical sector (same column) and the same disk (same row). Because logical sector 9,
For 10 and 11, the check data is the disk drive D0
The disk drive D0 is in use and cannot be multiplexed. In addition, logical sectors 5, 8, 1
Since 1 is being used for updating the check data (CD1) of logical sector 0, it cannot be multiplexed. As described above, even if the check data is arranged in the disk drives D0 to D3 while being shifted for each physical sector, it is rare that the data update in the sector unit can be performed in a multiplexed manner, and as a result, the performance of the data update in the sector unit is not improved. there were.

An object of the present invention is to store check data in a disk drive for check data, and when data update in sector units occurs, data update in sector units is performed in multiplex and data update is performed in multiplex. It is an object of the present invention to provide a disk array device that increases the number of logical sectors that can be executed and consequently speeds up data update processing in sector units.

[0008]

In order to achieve the above-mentioned object, the present invention provides a disk array control device in which data transfer to each disk drive and check data generation processing accompanying data update can be performed in a multiplexed manner. It has a buffer memory that stores the data transferred from the host computer and the data required to generate the check data for each disk drive, and each disk drive has a controller that controls the disk drive positioning and data transfer. As a disk drive for use, a storage means that enables high-speed access is provided. Further, a semiconductor disk having a backup function by a non-volatile mechanism is provided as a storage unit that enables high-speed access.

[0009]

According to the present invention, when a plurality of sector unit data updates occur, the write data transferred from the host computer is temporarily stored in the buffer memory means for each disk drive and the next input / output command can be accepted. It will be in a state. Then, when the subsequent command from the host computer is data update in units of sectors other than the same disk drive, the command is further accepted, and the write data transferred from the host computer is temporarily stored in the buffer memory means for each disk drive. . Further, in order to generate new parity check data of the check group of the updated logical sector, an instruction to read the old data and old check data of the updated logical sector is issued to each disk drive. Since the check data is stored in the external storage device that can be accessed at high speed, the reading is completed without positioning time. Since the user data is stored on the magnetic disk, positioning time is required when reading the old data of the updated logical sector.
When positioning is completed and the old data of the updated logical sector is read, new check data is generated from the update data, old data, and old check data stored in the buffer for each disk drive, and the check data disk is created. Issue a write command to the drive. It also issues an update data write command to the disk drive of the logical sector to be updated. As described above, since the check data is stored in the external storage device that can be accessed at high speed, the writing of the check data is completed without positioning time. Further, since the user data is stored on the magnetic disk, positioning time is required even when writing new data of the updated logical sector. When the positioning is completed and the writing of the new data of the updated logical sector is completed, the completion of the data update is reported to the host computer. In this way, the disk drive for check data is inactive for a longer period of time than the disk drive for storing user data. During this idle time, it is possible to process the update of the check data accompanying the data update of other logical sectors, and it is possible to multiplex the data update in units of a plurality of sectors.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows an embodiment of the present invention and shows the configuration of a disk array system using a semiconductor disk as a storage means capable of high speed access. This system includes a host computer 1, a disk array controller 9, and a plurality of disk drive controllers (four in this embodiment) 12a to 12d and disk drives (four in this embodiment) D0 connected to the controllers.
It is composed of D3. A semiconductor disk is used for the disk drive D3.

The disk array controller 9 is the host bus 2
Is connected to the host computer 1, and is connected to the disk drive controllers 12a to 12d via the disk bus 7. Each disk drive controller 12a-1
The disk drives D0 to D3 are connected to 2d, and the positioning with respect to the disk drives D0 to D3,
Controls data transfer. Disk drive D0-D3
Includes a check data disk drive D3 for storing check data and disk drives D0 to D2 for storing user data. Further, the disk drive controllers 12a to 12c have buffer memories 13a to 13c for buffering the read / write data of the corresponding disk drives D0 to D2.
It has c.

The disk array controller 9 has a plurality of buffer memories 10a to 10d corresponding to the respective disk drives D0 to D3. Buffer memory 10a
10d to 10d are connected to the disk bus 7 via a plurality of transceivers (data transmitting / receiving devices) 11a to 11d. The transceivers 11a to 11d execute data transfer with the host computer 1 and the disk drives D0 to D3. The buffer memories 10a to 10d store write data transferred from the host computer 1 and read data transferred from the disk drives D0 to D3. Further, since the buffer memories 10a to 10d enable the multiplex operation of updating the data, the buffer memories 10a to 10d can be controlled by being divided into a plurality of areas (here, three). further,
The disk array control device 9 uses the check data generation circuit 8
Have. The check data generation circuit 8 generates check data from the data stored in each of the buffer memories 10a to 10d and transfers it to the check disk drive D3 via the disk bus 7.

The disk array control device 9 also has a read transceiver 1 for transferring the data read from each of the disk drives D0 to D3 to the host computer 1.
1e, CP that controls the entire disk array controller 9
A U3, an interface 6 between the CPU 3 and the host computer, an interface 6 between the CPU 3 and the disk, and a memory 4 are provided. The interface 6 with the disk is used for each of the disk drive controllers 12a to 12d.
Is a circuit for transmitting commands and the like and receiving status and the like. The interface 5 with the host computer is an interface with the host computer 1, and is a circuit for receiving commands and the like from the host computer 1 and transmitting status and the like. The memory 4 consists of RAM and ROM, and C
The program of PU3 and various control data are stored.
Further, the allocation of physical sectors and logical sectors of the disk drives D0 to D3 of this disk array system is shown in FIG.
Is similar to.

Next, a case in which the data update processing of the logical sector 0 and the logical sector 10 is performed in a multiplex manner in the embodiment will be described with reference to FIGS. 1 and 2. FIG. 2 is a schematic diagram showing the relationship between the operation of the data update processing and time in this case. In FIG. 2, the shaded blocks are
Indicates the transfer time of the data indicated by DA0, DB0, etc., and
The blocks labeled with M1 to TM4 indicate the positioning time of the disc. First, a data update command for the logical sector 0 is received from the host computer 1, and the write data DA0 is stored in the area B00 of the buffer memory 10a. Then, from the host computer 1 to the logical sector 10
The data update command is received and the write data DB0 is stored in the area B10 of the buffer memory 10b. CPU
Reference numeral 3 denotes the old data DA2 of the logical sector 0 necessary for updating the check data accompanying the data update of the logical sector 0, and the old check data D of the check data group of the logical sector 0.
Read A1. First, the data DA of the logical sector 0 (physical sector 0) is written to the disk drive controller 12a.
A read instruction of 2 is issued. Since the disk drive D0 is a magnetic disk, positioning time occurs. The disk drive controller 12a completes positioning of the disk drive D0, and the logical sector 0 is stored in the buffer memory 13a.
Until data DA2 of (physical sector 0) is stored, C
The connection with the PU 3 is cut off and the disk bus 7 is opened. Subsequently, the CPU 3 causes the disk drive controller 12d to
An instruction to read the old check data (physical sector 0) DA1 of the check data group of logical sector 0 is issued. Since the disk drive D3 is a semiconductor disk, the old check data (physical sector 0) DA1 is read without positioning time. The read old check data DA1 is stored in the area B30 of the buffer memory 10d.

During the read positioning time of the disk drive D0, the CPU 3 checks the old data DB2 of the logical sector 10 and the old check of the check data group of the logical sector 10 necessary to update the check data accompanying the data update of the logical sector 10. The data DB1 is read. First,
The CPU 3 instructs the disk drive controller 12b to read the data DB2 of the logical sector 10 (physical sector 3). Since the disk drive D1 is a magnetic disk, positioning time is generated. The disk drive controller 12b and the CPU 3 continue until the positioning of the disk drive D1 is completed and the data DB2 of the logical sector 10 (physical sector 3) is stored in the buffer memory 13b. To disconnect the disk bus 7. Then, the disk drive controller 12d is instructed to read the old check data (physical sector 3) DB1 of the check data group of the logical sector 10. Disk drive D
Since 3 is a semiconductor disk, the old check data DB1 is read without positioning time. The read old check data DB1 is the area B31 of the buffer memory 10d.
Stored in.

When the positioning of the disk drive D0 is completed and the data DA2 of the logical sector 0 (physical sector 3) is stored in the buffer memory 13a, the CPU 3 is reconnected to the disk drive controller 12a and the old data of the logical sector 0 is reconnected. DA2 is stored in the area B01 of the read buffer memory 10a. The CPU 3 writes the write data (physical sector 0) stored in the area B00 of the buffer memory 10a to the disk drive controller 12a when the reading of the old data DA2 of the logical sector 0 is completed.
A write command of DA0 is issued. The disk drive controller 12a stores the write data DA0 in the buffer memory 13a and positions the disk drive D0. The disk drive controller 12a is
The connection with the CPU 3 is cut off and the disk bus 7 is opened until the disk drive D0 completes the positioning and the writing of the write data DA0 stored in the buffer memory 13a is completed.

Further, the CPU 3 uses the old data DA of the logical sector 0 stored in the area B01 of the buffer memory 10a.
2, the write data DA0 to the logical sector 0 stored in the area B00 of the buffer memory 10a, and the old check data DA1 stored in the area B30 of the buffer memory 10d are transferred to the check data generation circuit 8 and the new check data DA3 is transferred. To generate.

Then, the generated new check data DA3
Write command of the disk drive controller 12d
To publish. Since the disk drive D3 is a semiconductor disk, data transfer is performed without positioning, so new check data DA3 is transferred from the check data generation circuit and written to the disk drive D3.

During the write positioning time of the disk drive D0, the reading of the disk drive D1 is completed, and the old data DB2 of the logical sector 10 (physical sector 3) is stored in the buffer memory 13b. And CPU
3 is reconnected to the disk drive controller 12b and stores the old data DB2 of the logical sector 10 in the area B11 of the read buffer memory 10b. When the CPU 3 finishes reading the old data DB 2 of the logical sector 10,
In the disk drive controller 12b, the logical sector 10 stored in the area B10 of the buffer memory 10b
Issue a write command for write data DB0. The disk drive controller 12b stores the write data DB0 in the buffer 13b and positions the disk drive D1. The disk drive controller 12b uses the write data DB stored in the buffer memory 13b after the disk drive D1 has completed positioning.
The connection with the CPU 3 is cut off and the disk bus 7 is opened until the writing of 0 is completed.

Further, the CPU 3 uses the old data D of the logical sector 10 stored in the area B11 of the buffer memory 10b.
B2, write data DB0 of the logical sector 10 stored in the area B10 of the buffer memory 10b, and old check data DB1 stored in the area B31 of the buffer memory 10d are transferred to the check data generation circuit 8 and new check data DB3 is transferred. To generate. Then, the write command for the generated new check data DB3 is issued to the disk drive controller 12d. Disk drive D
Since 3 is a semiconductor disk, data transfer is performed without positioning, so new check data DB3 is transferred from the check data generation circuit and written to the disk drive D3.

At the time when the write positioning of each of the disk drives D0 and D1 is completed, the buffer memory 13a of the disk drive controller 12a is transferred to the disk drive D0.
b from the buffer memory 13b to the disk drive D1
Write data DA0 and DB0 are transferred to
Data update is completed. When the data update is completed,
CPU 3 is a disk drive controller 12a and 1
2b are reconnected to each other, and the end status is reported.

By the way, in the conventional array disk device shown in FIG. 4, the data of logical sector 0 is updated and the disk drive for storing the check data makes one round in the same area and the same physical sector (same column). Considering a logical sector other than the disk (in the same row) in which data can be updated in multiplex, only the logical sector 4 or 7 is used.
In the present embodiment, considering the same area in FIG. 3, it is possible to perform data update multiplex processing with any of the logical sectors 4, 7, 10 and with any of 5, 8, 11.

[0023]

As described in detail above, according to the present invention, even in a disk array system adopting a method of storing check data in a check data disk drive, when data update in a plurality of sector units occurs. It is possible to multiplex the data update in sector units and increase the number of logical sectors in which the multiple data update can be performed. Further, as a result, it is possible to speed up the data updating process in units of sectors.

[Brief description of drawings]

FIG. 1 is a block diagram showing a main part of a disk array control device according to an embodiment of the present invention.

FIG. 2 is a schematic diagram for explaining the operation and time relationship of the data update processing of the embodiment.

FIG. 3 is a block diagram for explaining a conventional disk array device.

FIG. 4 is a block diagram for explaining another conventional disk array device.

[Explanation of symbols]

 1 Host Computer 2 Host Bus 3 CPU 4 Memory 5 Host Interface 6 Disk Interface 7 Disk Bus 8 Check Data Generation Circuit 9 Disk Array Controller 10a-10d Buffer Memory 11a-11e Transceiver 12a-12d Disk Drive Controller 13a-13c Buffer Memory 14 Controller D0 to D3 disk drive

Claims (2)

[Claims] 1. A disk array controller for controlling a plurality of disk drives and controlling data transfer between each of the disk drives and a host computer, and a check of each write data recorded in each of the plurality of disk drives. In a disk array device that includes a check disk drive for storing data and checks data, the disk array control device can perform multiple data transfers to each disk drive and check data generation processing associated with data update. As described above, a buffer memory for storing the data transferred from the host computer and the data necessary for generating the check data is stored corresponding to the disk drive, and each of the disk drives has a controller for positioning the disk drive and controlling the data transfer. Equipment A disk array device characterized by comprising storage means that enables high-speed access as a check disk drive, shortening the check disk access time, and increasing the multiplicity of data update processing to each of the plurality of disk drives. . 2. The disk array device according to claim 1, wherein a semiconductor disk having a backup function by a non-volatile mechanism is provided as a storage means capable of high-speed access.