JPH10326158A - Storage array and array controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To attain a fast operation of a storage array via the parallel processings carried out between drive trains, by making use of a bus release period secured by a bus release/occupation function of a storage. SOLUTION: The array controller 3 of a disk array 1 has individual buffers 81a to 81c, 82a to 82c, 83a to 83c and 84a to 84c corresponding to the disk devices 10, which are connected every three pieces to the buses 41 to 44 respectively. In regard to the data read processing of the bus 41, the data, which are alternately transferred from three disks A1, B1 and C1 are transferred to the corresponding buffers 81a, 81b and 81c and accordingly arranged and stored there. In regard to the data read processing of the bus 41, the data to be written are previously stored in the buffers 81a, 81b and 81c and the corresponding data are transferred to one of disks A1, B1 and C1 that sent a data transfer request.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a storage device array in which a plurality of storage devices are controlled by an array controller, and responds to accesses from a main computer by imitating one storage device. .

[0002]

2. Description of the Related Art Conventionally, a disk array has been proposed as a storage device which enables high-speed access and guarantees high reliability. The disk array is systematized as a system that includes a plurality of small disk devices such as a hard disk device and enables high-speed access to a large disk device. This systematized method is RA
This is called an ID (Redundant Arrays of Inexpensive Disks), and the basic configuration from level 1 to level 5 is considered. Here, Level 3 related to the present invention
Will be described.

In the RAID of level 3, a plurality of disk devices including one parity disk device and two or more data disk devices are connected to different buses so that they can be accessed in parallel. . When data is stored in response to a request from the main computer, the data is divided into 1-byte units, and the data is dispersedly stored in each of the data disk devices, and a parity which is an exclusive OR of the divided data is stored. The data is calculated and stored in the parity disk device. That is, the data is distributed and stored in each of a plurality of disk devices connected to different buses.

As described above, since data that should be written to one disk device is written to a plurality of disk devices in a distributed manner, the frequency of reading and writing to and reading from one disk device is considered in total. Since the data amount is reduced, the processing speed is improved as a whole.

In addition, since parity data is stored as one of a plurality of disk devices as a parity disk device, even if data cannot be read from one disk device, the remaining disk devices can be read. , It is possible to generate data that is to be read from the one disk device based on the data from. As a result, data to be read can be generated even when a read error occurs in one disk device in the data read processing,
The reliability is improved.

By the way, for example, if a SCSI standard bus is used, a predetermined number of disk devices such as seven can be connected to one bus. Therefore, in order to realize a large-capacity disk array in addition to high speed and high reliability, a plurality of disk devices including the above-mentioned parity disk device are connected by connecting a plurality of disk devices to one bus. In general, a disk array provided with a plurality of disk device sets is generally used.

Here, the configuration of a conventional general disk array will be described in detail with reference to FIG. FIG. 5 is a block diagram showing a schematic configuration of a conventional disk array 100. As shown in FIG. The disk array 100 includes an array controller 30
It is composed of 0 and 12 disk devices 10.

The twelve disk units 10 are connected to four buses 41 to 44, respectively. In addition,
In order to distinguish the twelve disk devices 10, the symbols shown in FIG. 5 are used, and disks A1, disk A2, disk A3, disk A4, disk B1, disk B
2, disk B3, disk B4, disk C1, disk C2, disk C3, and disk C4.

In the example shown in FIG. 5, the disks A4, B4 and C4 connected to the bus 44 are parity disk devices. That is, the array controller 300 includes the disks A1 to A4, the disks B1 to B4,
Discs C1 to C4 are provided. Each set of four disk devices is connected to a different bus, that is, buses 41 to 4.
4, the array controller 300
Are controlled in parallel.

For example, in the data writing process, the data from the host device 500 is divided into three data by the bit operation circuit 50, and the parity data of the three data is calculated, and the disks A1 to A4 and the disks B1 To B4 or any one of disks C1 to C4
The data is stored in one of the disk devices via buses 41 to 44, respectively.

On the other hand, in the data reading process, disks A1 to A4, disks B1 to B4 or disks C1 to C4
Data from the corresponding four disk devices of C4 are read out to buffers 31 to 34, combined by the bit operation circuit 50, and transferred to the host device 500. If error correction is necessary, the bit operation circuit 50 performs error correction using the parity data in the buffer 34.

As described above, the disk array 100 achieves high speed by accessing disk devices connected to different buses in parallel by the array controller 300. Since the disk array 100 includes four independent access systems using different buses, the disk array 100 is referred to as a system array 4 disk array. A set of a plurality of disk devices operating in parallel is called a drive row. Therefore, disk array 1
00 can be said to be a disk array of the drive row 3. Hereinafter, the discs A1 to A4 will be referred to as an A drive row,
B4 to B4, and a set of disks C1 to C4 is a C drive. A disk device in a general disk array can be expressed by a matrix structure such as a system row n (n is a natural number; n ≧ 2) and a drive row m (m is a natural number; m ≧ 2). It can be said that the processing speed is improved by the parallel processing.

In general, parallel processing between drive trains cannot be performed. This is because, for example, when the disk A1 is accessed in FIG. 5, the bus 41 is occupied by the disk A1, and the disks B1 and C1 cannot be accessed. However, for example, in the SCSI standard, a disk device connected to the same bus can be accessed in parallel using a function called disconnect / reconnect. This disconnect / reconnect function is performed by the S connected to the bus 41 in FIG.
CSI controller 21, disk A1, disk B1
And the disk C1.

For example, when the SCSI controller 21 issues a command for requesting data reading from the disk A1, the disk A1 receives this command and executes processing based on the command. That is, the disk A1 reads data from the disk surface and transfers the read data to the SCSI controller 21.

At this time, if the disk A1 has a disconnect / reconnect function, the disk A1
During an execution period of a process based on a command from the CSI controller 21, for example, during a period when the bus 41 is not required, such as a period during which data is read from the disk surface, the connection with the SCSI controller 21 is temporarily disconnected (disconnected) and the bus 41 is disconnected. During the period when the bus 41 is required.
The bus 41 is occupied by reconnection (reconnect) with the SI controller 21.

Since the disk device having the disconnect / reconnect function does not occupy an unnecessary bus, the array controller uses the bus release period to access the disk devices connected to the same bus in parallel. It is possible to do. For example, the SCSI controller 21 can transmit a data read request command to the disk B1 or the disk C1 after issuing a data read request command to the disk A1, waiting for the disk A1 to be disconnected.

Therefore, in addition to the parallel processing between the system rows as described above, the parallel processing between the drive rows, for example, the parallel processing between the drive rows A to C in FIG. It was thought that high speed could be realized.

[0018]

However, when the disk devices connected to the same bus are accessed in parallel by using the disconnect / reconnect function as described above, for example, each disk may be accessed in response to a data read request command. The data transferred from the device is not always transferred in the order of the data read request, and when the data is transferred in a divided manner, it may be transferred as a set of data for each disk device. Not exclusively.

For example, the disk array 100 shown in FIG.
Even if the SCSI controller 21 issues the data read request command in the order of the disk A1, the disk B1, and the disk C1, the bus is released, so if the disk C1 completes the data read first, the disk C
The data from 1 is transferred first. Further, in the case where the disk C1 transfers the data to be read by dividing it into two parts, if the bus is released after the first divided data is transferred, the disk C1 is disconnected from the disk A1 or the disk B1. Data may be transferred.
In that case, the data from disk C1 is
Alternatively, the data is divided by the data from the disk B1.

Therefore, for example, data transferred from the disks A1, B1 and C1 connected to the bus 41 are stored in the buffer 31 in a mixed state as shown in FIG. In FIG.
When the array controller 300 reads data in parallel from the A drive row, the B drive row, and the C drive row,
Since the above-described situation occurs in each of the four buses 41 to 44, as shown in FIG.
The data stored in the buffers 34 to 34 cannot be handled among the buffers 31 to 34.

However, in the disk array 100, the data is distributed and stored in any one of the four disk drives A, B, and C, and the data read from the four disk drives are respectively stored. The distributed processing is performed between the system strings, which is combined and output to the host device. Therefore, the data transferred in the order as shown in FIG. 7A is combined as data for each disk device as shown in FIG. 7B, and between the system rows, that is, buses 31 to 34. It was necessary to rearrange as data that was taken between. Therefore, the time shortened by the parallel processing between the drive arrays is offset by the time required for data reduction. Therefore, it has been difficult to realize a high-speed disk array by parallel processing between drive arrays.

An object of the present invention is to realize a high-speed storage device array by parallel processing between drive rows by shortening the time required for data operations such as data arrangement at the time of data reading.

[0023]

According to the first aspect of the present invention, there is provided a storage device array having a plurality of buses, each of which has a plurality of storage devices. By connecting the devices, a plurality of drive columns, which are a set of storage devices connected to different buses and operating in parallel, are provided, and are configured to be accessed by the array controller in drive column units. A bus occupancy / release function that releases a bus during a period when a bus is not required, and occupies a bus during a period when a bus is required, at the time of executing processing based on a command for at least one of reading and writing of the data. In the storage device array provided with the storage device, individual storage areas corresponding to the storage devices are provided, and based on a request from the host device, the read target or A command issuing means for issuing a processing request command using the bus release period to each storage device in the drive array to be written, and a data transfer request from a storage device executing a process based on the processing request command. In the case, the requesting device specifying means for specifying the storage device that has requested the transfer, the storage area specifying means for specifying the storage area corresponding to the storage device specified by the requesting device specifying means, and a request from the host device for reading data When the request is a request, the corresponding data is transferred from the storage device specified by the requesting device specifying unit to the storage area specified by the storage region specifying unit. On the other hand, when the request from the host device is a data write request. Corresponds to the storage device specified by the requesting device specifying means from the storage area specified by the storage area specifying means. Transfer execution means for transferring data, and when the request from the host device is a data read request, and when there is a drive train for which transfer has been completed, the data is stored in a storage area corresponding to each storage device of the drive train. Generates the data to be read from the data
On the other hand, when the request from the host device is a data write request, and when the data to be written is transmitted from the host device, the data to be written is stored in the storage area corresponding to each storage device in the drive array to be written. And data calculation means for storing the data in a distributed manner.

The present storage device array has n (n is a natural number; n)
If m (m is a natural number; m ≧ 2) storage devices are connected to each of the ≧ 2) buses, the (m × n) storage devices are connected to n (n) connected to different buses. It is divided into sets of storage devices and controlled as m rows of drive rows.

At this time, since the n storage devices constituting the drive train are connected to different buses,
The data read processing and the data write processing are speeded up by distributing and storing the data stored in one storage device in the n storage devices.

Each storage device connected to the bus has a bus release / occupation function. When a command requesting at least one of data writing and reading is received, processing based on the command is performed. Run. At that time, unnecessary bus occupation is not performed. That is, during the execution period of the processing based on the command, the bus is released during a period when the bus is not required, for example, a period during which data is read from the disk surface, and the bus is released during a period during which the read data is transferred, for example, during the period when the read data is transferred. Occupies the bus.

Therefore, parallel processing for a plurality of storage devices connected to one bus can be performed by utilizing the bus release period by the bus release / occupation function. Therefore, using the bus release period, the command issuing unit requests a plurality of drive arrays to perform processing in parallel based on a processing request from the host device. That is, a processing request command is issued to each storage device in a plurality of drive arrays.

For example, when there are two or more processing requests from the host device, first, a processing request command is issued to each storage device of the corresponding one drive row. As a result, each storage device of the drive row executes processing based on the processing request command, and releases the bus when the bus is not required by the bus release / occupation function. Utilizing the bus release period, the command issuing unit issues a processing request command to each storage device of another corresponding drive row.

The present storage device array has storage areas individually provided corresponding to the respective storage devices. Each storage device that is executing the processing based on the processing request command by releasing the bus reads, for example, a predetermined amount of data from the disk surface, or, for example, prepares to write the data, and then occupies the bus again to occupy the data. When a transfer request is made, the requesting device specifying unit determines which storage device connected to the bus is the transfer request in response to the data transfer request, and makes a transfer request. Identify the storage device. Further, the storage area specifying means specifies storage areas individually provided corresponding to the storage devices specified by the requesting apparatus specifying means.

When the storage device and the storage area are specified by the requesting device specifying unit and the storage area specifying unit, the transfer executing unit, if the request from the host device is a data read request, specifies the specified storage device. Is used as a transfer source, and data transfer is performed using the specified storage area as a transfer destination. On the other hand, if the request from the host device is a data write request, data transfer is performed with the specified storage area as a transfer source and the specified storage device as a transfer destination.

In the case where the request from the host device is a data read request and the data transfer on a drive-row basis is completed, the data operation means is provided with an individual memory corresponding to each storage device of the drive row on which the transfer has been completed. When the data to be read is generated from the data stored in the storage area of the storage device, and the request from the host device is a data write request, and when the data to be written is transferred from the host device, the corresponding drive train The data to be written is distributed and stored in storage areas corresponding to the respective storage devices.

As described above, in the data reading process, when data is read in parallel from a plurality of drive rows using the bus release period by the bus release / occupation function of the storage device, a buffer or the like provided for each bus is used. In this storage area, data transferred from a plurality of storage devices is stored as it is, so that the data does not become a set of data for each storage device as shown in FIG. In addition, since such a situation occurs for each bus, when data read for each drive row is generated as data to be read, correspondence can be taken between the buses as shown in FIG. It was necessary to rearrange the data as a set of data for each storage device. As a result, it has been difficult to speed up the data reading process by the parallel processing between the drive arrays due to the time required for the data operation for the organization.

On the other hand, the present storage device array has individual storage areas corresponding to the respective storage devices. In the data reading process, when there is a data transfer request from the storage device, the requesting device identification unit
Determine which storage device the data transfer request is from,
The storage area specifying means specifies a storage area provided corresponding to the storage device requesting the transfer. Thus, when the data transfer is executed by the transfer execution unit, the data transferred from each storage device is stored in the individual storage area corresponding to each storage device.

As a result, regardless of whether data is transferred alternately from a plurality of storage devices connected to one bus or in any order. The data from each storage device is organized and stored as a set of data for each storage device. That is, the transferred data is directly arranged as shown in FIG.

Therefore, when the data generation means determines the end of the data transfer for each drive row, it can immediately generate the data to be read based on the data stored in the corresponding storage area. This eliminates the need to temporarily store the transferred data and then store the data again as data for each storage device, and the data is organized at the same time as the data is transferred from the storage device, which is necessary for data organization. The time can be reduced, and the speed of the data reading process can be increased by the parallel processing between the drive rows.

On the other hand, in the data writing process, when data is written in parallel to a plurality of drive rows using the bus release period by the bus release / occupation function of the storage device, the data is stored in a buffer or the like provided for each bus. In the area, the data to be written transmitted from the host device is stored in the order of transmission. Therefore, when a transfer request for writing is alternately made from a plurality of storage devices, it is necessary to perform a data operation for selecting and transferring data corresponding to the transfer request from a series of data stored in the buffer. Come. In addition, since such data manipulation is required for each bus, it is difficult to speed up data writing processing by parallel processing between drive trains due to the time required for the data manipulation.

On the other hand, the present storage device array has individual storage areas respectively corresponding to the respective storage devices, and the data operation means transfers the host data to the storage regions corresponding to the respective storage devices to be written. The data to be written transmitted from the device is distributed and stored.

Then, when there is a data transfer request from each storage device to be written, the requesting device specifying means specifies the storage device that has requested the transfer from among the plurality of storage devices. Further, the storage area specifying means specifies a storage area corresponding to the storage device. As described above, since the write target data from the host device is stored in advance in the storage area corresponding to the write target storage device requesting the data transfer, the transfer execution unit requests the transfer. The corresponding data can be immediately transferred from the specified storage area to the storage device.

That is, since the data to be written is stored in the individual storage areas in advance, the storage area corresponding to the storage device that has requested the transfer is specified and the data is transferred. As a result, regardless of which storage device among the plurality of storage devices requests the data transfer, and in any order, the corresponding data is immediately read. It becomes possible to transfer to those storage devices.

As a result, data for a plurality of storage devices connected to one bus is stored in a storage area such as a buffer, and data corresponding to the storage device requested to be transferred is stored in the data. This eliminates the waste of selecting and transferring data, and can reduce the time required for data operation. As a result, it is possible to realize high-speed data write processing by parallel processing between drive arrays.

The storage device array operates based on at least one of a data read request and a data write request from the host device. That is,
The storage device array may perform only data read processing, may perform only data write processing, or may perform both processing. Conceivable. However, a storage device array that performs both data reading and writing processes is generally used.

The storage device array apparently operates as one storage device from the host device. Normal,
A series of cluster numbers are set in the storage areas of a plurality of storage devices included in the storage device array. Data read and write requests from the host device are generally specified by the cluster number. Therefore, the storage device array stores the correspondence between the unit storage area for management managed by the cluster number and the physical storage area of the storage device, and is designated by the host device based on the correspondence. The storage area corresponding to the cluster number thus accessed is accessed.

In general, the host device continuously stores a series of data in a management storage area. For example, certain data is often stored in a continuous administrative storage area such as clusters “3”, “4”, and “5”. Therefore, in addition to the configuration described in claim 1 above,
It is conceivable to configure a storage device array as described in claim 2. That is, it is conceivable that a set of unit storage areas at the same position in each storage device constituting a drive array is used as a management unit storage area from the host device side, and the management unit storage areas are arranged by striping. .

Here, the unit storage area in the storage device refers to a storage area that is a unit for expressing the storage area of the storage device, such as a block or a sector. The management unit storage area from the host device side is, for example, the cluster described above.

Data is distributed and stored in each storage device of the drive array. At this time, for example, data is distributed and stored in the same block of each storage device. Therefore, the same set of blocks of each storage device constituting the drive array is used as a cluster. In the present storage device array, for example, a management unit storage area such as a cluster is allocated to a drive row using a technique called striping.

For example, the condition is as shown in FIG.
In FIG. 2, three storage devices are connected to the four buses, respectively, to form three rows of drives composed of four storage devices. In FIG. 2, A drive row, B
Drive row and C drive row. At this time, each storage device has storage areas of blocks “0” to “k”, and four storage areas that are the same block are collectively formed as a cluster for each drive row. This cluster is allocated to each drive row by a method called striping. Here, cluster “0” is stored in four storage areas of block “0” in drive A row, cluster “1” is stored in four storage areas of block “0” in drive B row, and block “0” is stored in drive C row. "In the four storage areas,
Block “3” is stored in the storage area of block “1” in drive A, cluster “4” is stored in block “1” of drive B, and cluster “5” is stored in block “1” of drive C. k ”.

As described above, the host device generally stores a series of data in a continuous storage area for management. For example, a series of data is divided into clusters “3”, “4”,
In the case where data is stored in a continuous management unit storage area such as “5”, in FIG.
The data is stored separately in the B drive row and the C drive row. As a result, the time required for writing and reading data can be reduced by the parallel processing between the drive rows. That is, the present storage device array is advantageous for sequential access from the host device.

For example, the reading process when the data is video data or the like is a sequential reading process as described above. Therefore, assuming a host device that performs such a reading process, it is desirable to further configure the configuration as described in claim 3. That is, claim 2
In addition to the configuration shown in the above, when there is a data read request from the host device, a predetermined number of management unit storage areas continuously arranged after the management unit storage area specified by the read request It is desirable to be configured to read data.

In this case, data is read from a predetermined number of management unit storage areas continuously arranged after the management unit storage area such as a designated cluster. For example, when the host device requests reading of data stored in the cluster “1”, the management unit storage areas continuously arranged after the cluster “1” are cluster “2”, cluster “3”, Cluster “4” is obtained. At this time, for example, if the predetermined number is 3, cluster “1” and cluster “2”
And read data from cluster “3”, and the predetermined number is 2
If so, the data is read from the cluster “1” and the cluster “2”.

That is, a series of data such as movie data is generally continuously stored in a cluster which is a unit storage area for management. In this case, a read process for reproduction is performed by continuously performing clusters. Will be read. Therefore, when a read request from the host device is issued for one cluster, the time required for the data read process can be further reduced by pre-reading data from a cluster that is continuous with the cluster.

By the way, for example, in the SCSI standard, up to seven storage devices can usually be connected to one bus. Therefore, it is conceivable to increase the number of drive rows in the storage device array. In consideration of such an increase in the number of drive rows, a configuration according to a fourth aspect is preferable. That is,
In addition to the configurations described in the first to third aspects, it is preferable that the individual storage areas respectively corresponding to the storage devices are configured to be able to be added in accordance with the number of connected storage devices.

In this case, for example, a configuration is conceivable in which the storage area can be expanded by adding a memory module or the like. That is, when the number of drive rows increases, by adding this memory module, it is possible to secure storage areas corresponding to the respective storage devices. In addition to physically increasing the storage area, a predetermined storage area is provided in advance, and the storage area is divided according to the number of connected storage apparatuses, and individual storage areas corresponding to the respective storage apparatuses are respectively provided. It is also conceivable to secure an area.

As a result, the number of drive rows can be increased, and a scalable storage device array can be realized. In particular, when a storage area can be added with a memory module or the like, the required number of memory modules can be purchased according to the number of drive rows, which is economical. This is advantageous in that a sufficient storage area can be secured by purchasing a memory module.

The above has been described as an invention of a storage device array including a plurality of storage devices for releasing and occupying a bus, and an array controller. On the other hand, the present invention can be realized as an invention of an array controller which is enabled by connecting a storage device having a bus release / occupancy function as described above.

According to a fifth aspect of the present invention, there is provided an array controller, wherein a plurality of storage devices are connected to a plurality of buses, respectively, so that a drive row unit is a set of storage devices connected to different buses and operating in parallel. In the array controller configured to access in the, provided with a separate storage area corresponding to each storage device, based on a request from the host device, for each storage device of the drive row of the read target or write target, A command issuing unit that issues a processing request command using the bus release period, and when a data transfer request is issued from a storage device that is executing a process based on the processing request command, specifies the storage device that has requested the transfer. Request device specifying means, and storage for specifying a storage area corresponding to the storage device specified by the request device specifying means When the request from the host device is a data read request, the data is transferred from the storage device specified by the requesting device specifying device to the storage area specified by the storage area specifying device. When the request from the host device is a data write request, transfer execution means for transferring the relevant data from the storage area specified by the storage area specifying means to the storage device specified by the requesting device specifying means; Is a data read request, and if there is a drive row for which transfer has been completed, data to be read is generated from data stored in a storage area corresponding to each storage device in the drive row. , When the request from the host device is a data write request, or when data to be written is transmitted from the host device It is characterized in that and a data calculating means for dispersing and storing the data to be written into the storage area corresponding to each storage device of the drive train to be written.

Five means in the present array controller, namely, a command issuing means, a requesting device specifying means,
The operations of the storage area specifying means, the transfer executing means and the data calculating means are the same as those described above for the storage device array in claim 1, and therefore the description thereof is omitted here. This array controller is connected to the host device, and the
By connecting a plurality of storage devices having an occupation function, the same effect as the effect described in claim 1 is exhibited. In other words, in at least one of the data read processing and the data write processing processing in units of drive arrays, it is possible to realize high-speed processing by parallel processing between drive arrays.

Further, the present array controller is characterized in that:
In the same manner as described above, the data is pre-read based on the data read request from the host device, or the storage area corresponding to each storage device can be expanded in the same manner as described in claim 4. Of course, it is also possible.

[0058]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of a disk array 1 according to the present embodiment, which embodies a storage device array.

The disk array 1 of the present embodiment is connected to the host device 500, and operates as if it were a single storage device with respect to the host device 500 based on a request from the host device 500. As shown in FIG. 1, the disk array 1 includes an array controller 3 and twelve disk devices 10. The disk device 10 is a so-called hard disk device.

The twelve disk units 10 are connected to four buses 41 to 44, respectively. Less than,
In order to distinguish the twelve disk devices 10, the symbols shown in FIG. 1 are used to indicate disks A1, A2, A3, A4, B1, B2, B3, B4, C1, C.
2, disk C3, and disk C4.

The array controller 3 has a CPU 97 as control means and a ROM 9 as program storage means.
A RAM 99 serving as a temporary storage unit serving as a work area when a process based on a program stored in the ROM 98 is executed by the CPU 97;
4, a PCI bridge 63 for connecting the host device 500 via the host device 500, a cache memory 60 for transferring data to and from the host device 500, and a bit operation circuit 50 as "data operation means" for distributing and generating data. And

The array controller 3 has a configuration for independently controlling the four buses 41 to 44. That is, it has four independent access systems. Since each of these access systems has the same configuration, the configuration of one access system for controlling the three disks A1, B1, and C1 connected to the bus 41 will be described. Disk A1, B1, C1
A1, B1, C1
1, a buffer 81a as an individual storage area corresponding to the disk A1, a buffer 81b as an individual storage area corresponding to the disk B1, and an individual storage area corresponding to the disk C1. (Direct Memory Access Controller) 91 that transfers data between the SCSI controller 21 and the buffers 81a to 81c.
And a switch 7 for switching any one of the buffers 81a, 81b or 81c as a data transfer target buffer prior to data transfer by the DMAC 91.
1 is provided. As shown by broken lines in FIG. 1, each system is configured so that buffers 81d to 84d can be added. Note that, as described above, the other systems have the same hardware configuration, and a description thereof will be omitted. The first digit of the reference numeral of the component of each access system is given as “1” to “4” corresponding to each system.

The bit operation circuit 50, switches 71 to 74 of each access system, DMACs 91 to 94, S
CSI controllers 21 to 24 and PCI bridge 63
Are controlled by the CPU 97, but the control bus from the CPU 97 is not shown in FIG. 1 to avoid complexity. The SCSI controllers 21 and 2
4 and DMACs 91 to 94 correspond to “transfer executing means”.

The array controller 3 controls a set of four disk devices connected to different buses 41 to 44 as a drive train by independently controlling the four access systems as described above. . FIG.
In the table, a set of disk devices of the disks A1 to A4 is a drive array A, a set of disk devices of the disks B1 to B4 is a drive array B, and a set of disk devices of the disks C1 to C4 is C.
Drive rows are provided, and data is read and written in drive row units A to C. That is, the data is distributed and stored in the four disk devices 10 constituting the drive array in units of the drive arrays A to C.

As described above, the data to be written to one disk device is distributed and stored in the four disk devices 10 connected to the buses 41 to 44, respectively. The frequency of reading / writing and the amount of data to be read from / to 10 are reduced, and the overall processing speed is improved. This is called parallel processing between system rows.

Also, one of the disk arrays 1 of the present embodiment
The storage areas of the two disk devices 10 are managed as shown in FIG. FIG. 2 is an explanatory diagram showing the storage area of each storage device. The storage area of each disk device 10 is represented by a block address, which is an access unit, and includes storage areas from block “0” to block “k”. In the present embodiment, four storage areas in which the same blocks are put together for each drive row are defined as clusters. The host device 500 requests the disk array 1 to read or write data by designating this cluster. The disk array 1 has a correspondence between a cluster, a drive row, and a block address in advance, and accesses a corresponding block address of each disk device 10 in a corresponding drive row based on the correspondence.

Further, clusters are sequentially allocated to drive rows A to C by using a technique called striping. In the present embodiment, as shown in FIG. 2, the cluster “0” is stored in four storage areas of the block “0” in the A drive row.
And cluster “1” in four storage areas of block “0” in the B drive row, and 4
Similarly, cluster “2” is stored in one storage area, cluster “3” is stored in the storage area of block “1” of drive A row, cluster “4” is stored in block “1” of drive B row, and block “C” of drive C row Blocks “k” are sequentially assigned to “1” in the order of cluster “5”.

Further, the twelve disk devices 10 of the disk array 1 according to the present embodiment have a bus release / occupation function called disconnect / connect in the SCSI standard, and execute data read and write processing. Avoid unnecessary bus occupation during the period.

For example, the disk A connected to the bus 41
The function of this disconnect / reconnect will be described by taking 1 as an example. When the SCSI controller 21 accesses the disk A1 based on a data read and write request from the host device 500, a state in which data transfer with the disk A1 is possible (hereinafter referred to as “combining”) is performed, and a processing request is sent to the disk A1. Issue a command. When the disk A1 receives the processing request command from the SCSI controller 21, the disk A1 executes processing based on the command, but during the processing execution period, for example, during a period when the bus 41 is not required such as a period during which data is read from the disk surface. Disconnects the connection with the SCSI controller 21 once (disconnect) to release the bus 41, and reconnects with the SCSI controller 21 (reconnect) to occupy the bus when the bus 41 is required.

Thus, when the disk A1 is accessed via the bus 41, the bus 41 is not occupied continuously from the start to the end of the processing of the disk A1. Therefore, during execution of the processing of the disk A1, it is possible to access the disks B1 and C1 in parallel using the release period of the bus 41.

That is, each of the SCSI controllers 21 to 24 uses this disconnect / reconnect function.
Processing request commands can be issued in parallel to the three disk devices 10 connected to the buses 41 to 44, respectively. That is, data can be read from and written to the three drive rows A to C in parallel.

Hereinafter, the parallel processing between the drive rows, which is a feature of the disk array 1 of the present embodiment, will be described with reference to the flowcharts of FIGS. FIG. 3 is a flowchart of the command issuing process. This process is executed by the CPU 97 of the array controller 3 every time an access from the host device 500 occurs, based on a program stored in the ROM 98.

First, in the first step S1000, it is determined whether there is a data read or data write request from the host device 500. If there is no request from the host device 500 (S1000: NO),
This command issue processing ends. On the other hand, the host device 50
If there is a request from 0 (S1000: YES), S
Move to 1010.

In S1010, a request from the host device 500 is fetched. Then, in S1020, based on the cluster number specified by the host device 500, a drive row to be read or written is specified. For example, the A drive row is specified from the A to C drive rows.

In S1030, it is determined whether the request from the host device 500 is a read processing request.
Here, when it is determined that the request is a read processing request (S1
030: YES), and proceeds to S1040. On the other hand, when it is determined that the request is not a read processing request (S1030: N
O), that is, if it is a write processing request, S
Move to 1050.

If a positive determination is made in S1030, that is, when a data read request is made, the process proceeds to S10.
In 40, a read request command is issued to each disk device 10 in the drive row specified in S1020. For example, when the A drive row is specified in S1020, a read request command is issued to the four disks A1 to A4.
After executing the processing in S1040, the command issuing processing ends.

On the other hand, if a negative determination is made in S1030,
That is, in S1050 to which the process proceeds when a data write request is made, the bit operation circuit 50
From 00, the write target data transferred to the cache memory 60 via the PCI bridge 63 is divided. Here, the bit operation circuit 50 divides the data to be written into four data including redundant data.

In S1060, the data divided by the bit operation circuit 50 in S1050 is stored in the four buffers corresponding to each disk device 10 in the drive row specified in S1020. For example, when the A drive row is specified in S1020, the data is stored in the four buffers 81a to 84a corresponding to the disks A1 to A4.

In S1070, a write request command is issued to each disk device 10 in the drive row specified in S1020. For example, when the A drive row is specified in S1020, a write request command is issued to the four disks A1 to A4. After execution of the process of S1070, the command issuing process ends.

In S1040 and S1070, a read or write request command is issued using the release period of the buses 41 to 44. Therefore, in this command issuance processing, when requests from the host device are continuous, a read or write request command can be issued to a plurality of drive rows in parallel, and the drive rows A to C can be issued in parallel. Can be accessed. Also, CP
U97 corresponds to “command issuing means”, and the processing of S1040 and S1070 corresponds to processing as command issuing means.

Then, the disk device 10 which has received the read request command, for example, the disk device 10 connected to the bus 41, releases the bus 41 by disconnection and performs data read processing, and a predetermined amount of data is read. Is reconnected, the bus 41 is occupied and the transfer of the read data is requested. On the other hand, the disk device 10 that has received the write request command, for example, the disk device 10 connected to the bus 41, releases the bus 41 by disconnecting and prepares for data writing. Requests the transfer of data to be occupied and written.

Next, the data transfer processing will be described with reference to the flowchart of FIG. This process is an interrupt process executed when there is the above-mentioned data transfer request from the disk device 10 which is executing the process based on the process request command issued in the above-mentioned command issuing process. Note that this processing is also performed by R
CPU 9 based on a program stored in advance in OM98
7 is performed. Here, the description will be made on the assumption that a data transfer request is made from the disk A1.

First, in the first step S2000, the request is taken out from the disk device 10 which has made the transfer request. Then, based on this request, S201
In the case of 0, the disk device 10 which has requested the transfer is specified. Here, as described above, the disk device 10 is specified as the disk A1.

At S2020, the buffer corresponding to the disk A1 specified at S2010 is specified. That is, the buffer 81a is specified. S20
At 30, it is determined whether or not the data transfer request from the disk A1 is a transfer request for a read process. Here, if the transfer request is for a read process (S203)
0: YES), and proceeds to S2040. On the other hand, if the request is not a transfer request related to a read process (S2030: N
O), that is, if the transfer request is for a write process, the flow shifts to S2100.

If the determination is affirmative in S2030, that is, if there is a transfer request related to the read process, the flow advances to S2040 to instruct the DMAC 91 connected to the SCSI controller 21 to prepare for data transfer. Based on this instruction, the DMAC 91 makes preparations for transferring data from the SCSI controller 21 to the buffer 81a when there is a transfer request from the SCSI controller 21.

In the subsequent S2050, an instruction is issued to execute data transfer to the SCSI controller 21 connected to the disk A1 which has requested the transfer. Based on this instruction, SC
The SI controller 21 sends the request to the disk A1
, And issues a transfer request to the DMAC 91 to transfer the data to the DMAC 91. Since the DMAC 91 is preparing for the transfer in S2040, the DMAC 91 transfers the data from the SCSI controller 21 to the buffer 81a. This buffer 81a is the buffer specified in S2020 described above,
Switch 1 and transfer.

As a result, the corresponding data is transferred from the disk device 10 that has requested the transfer to the individual buffer corresponding to the disk device 10. S2
In step 060, it is determined whether the transfer is completed for each drive row. When the transfer from the disk A1 is performed as described above, it is determined whether or not all the data transfers from the drive A row have been completed. Here, when it is determined that the transfer from the A drive row has been completed (S2060: YES), S20
Move to 70. On the other hand, if it is determined that the transfer from the drive A row has not been completed (S2060: NO), the data transfer processing ends.

In S2070, the data stored in the buffers 81a to 84a storing the data transferred from the disks A1 to A4 in the A drive row is read.
In subsequent S2080, the bit operation circuit 50 sets the buffer 8
Data to be read is generated based on the data read from 1a to 84a. At this time, if error correction is necessary, error correction is performed using redundant data. And
The restored data is stored in the cache memory 60.

In S2090, the completion of the data read process for the A drive row is reported to the host device 500. After performing the process of S2090, the data transfer process ends. In response to the completion report of the data read process in S2090, the host device 500 reads the data to be read from the cache memory 60.

If a negative determination is made in S2030, that is, if there is a transfer request related to the write processing, the process proceeds to S2100, where the connection to the SCSI controller 21 is made in order to transfer the data to be written to the disk A1 for which the transfer was requested. The DMAC 91 is instructed to prepare for data transfer. Based on this instruction, the DMAC 91 prepares to transfer the data stored in advance in the buffer 81a to the SCSI controller 21 when there is a transfer request from the SCSI controller 21.

At S2110, the CPU instructs the SCSI controller 21 connected to the disk A1 which has requested the transfer to execute the data transfer. Based on this instruction, SC
The SI controller 21 instructs the DMAC 91 to transfer data. Since the DMAC 91 is preparing for data transfer in S2100, the DMAC 91 transfers the corresponding data from the buffer 81a to the SCSI controller 21. The SCSI controller 21 takes in the data transferred from the DMAC 91 and transfers it to the disk A1 that has requested the transfer. The buffer 81a is switched and selected by the switch 71. The data stored in advance in the buffer 81a is the data stored in S1060 in FIG.

As a result, the corresponding data can be immediately transferred to the disk device 10 that has requested the transfer. In S2120, it is determined whether the transfer in the drive array is completed. When the transfer to the disk A1 is performed as described above, it is determined whether or not all the data transfer to the drive A row has been completed. Here, if it is determined that the transfer to the drive A row has been completed (S2120: YES), S2
Move to 130. On the other hand, if it is determined that the transfer from the drive A row has not been completed (S2120: NO),
This data transfer processing ends.

In S2130, the completion of the data write processing to the A drive row is reported to the host device 500. After performing the process of S2130, the data transfer process ends. Although the description here has been made on the assumption that a data transfer request has been made from the disk A1, the same applies to a case where a data transfer request has been made from another disk device 10. Further, the CPU 97 corresponds to “requesting device specifying means” and “storage area specifying means”, S2010 corresponds to processing as requesting device specifying means, and S2020 corresponds to processing as storage area specifying means.

Next, the effects exhibited by the disk array 1 of the present embodiment will be described. The effect in the data reading process will be described first, and then the effect in the data writing process will be described. In addition, in order to facilitate understanding of the description, description will be made in comparison with the above-described conventional problems.

In parallel data read processing from a plurality of drive rows, the conventional disk array 100
As described above with reference to FIG. 5, the data transferred from the three disk devices 10 connected to the buses 41 to 44 are stored as they are in the buffers 31 to 34 provided for the buses 41 to 44, respectively. Will be. That is, since one buffer 31 is connected to one bus 41, the three disks A connected to the one bus 41
The data transferred from 1, B1, and C1 is stored in the one buffer 31 as it is. Therefore, the data is not collected for each of the disks A1, B1, and C1 (see FIG. 6). Moreover, each of the four buses 4
Since such a situation occurs every 1 to 44 (FIG. 7)
7A), when the data read for each drive row is generated as the data to be read, the buses 41 to 44 as shown in FIG. It was necessary to rearrange it as a set of data. As a result, it has been difficult to speed up the data reading process by the parallel processing between the drive arrays due to the time required for the data operation for the organization.

On the other hand, in the disk array 1 of the present embodiment, as shown in FIG. 1, the buses 41 to 44 correspond to the three disk devices 10 connected to the buses 41 to 44, respectively. Three buffers 81a-
81c, 82a to 82c, 83a to 83c, 84a to 8
4c. For example, looking at bus 41, 1
Three disks A1, B1,
Each individual buffer 81a, 81b,
81c.

When there is a data transfer request from the disk device 10, the data transfer process described with reference to FIG. 4 is executed to identify the disk device 10 that has requested the transfer (S2010 in FIG. 4). Then, the buffer corresponding to the disk device 10 which has requested the transfer is specified (S2020 in FIG. 4). For example, it specifies that the disk device 10 that has requested the transfer is the disk A1, and further specifies the individual buffer 81a corresponding to the disk A1.

Thus, the disk A1 and the buffer 81a
Is specified, the corresponding data is transferred from the disk A1 to the buffer 81a by switching the switch 71 (S2040 and S2050 in FIG. 4). As a result, the data to be transferred is stored in individual buffers corresponding to the respective disk devices 10. For example,
As for the bus 41, data of the disk A1 is sent to the buffer 81a, and data of the disk B1 is sent to the buffer 81b.
And the data from the disk C1 is stored in the buffer 81c.
Respectively.

As a result, for example, the 3 connected to the bus 41
Even if the data transfer requests are alternately received from the three disks A1, B1, and C1, and in any order, the three disks A1, B1, and The data from C1 is organized and stored as individual data in individual buffers 81a, 81b and 81c corresponding to the disks A1, B1 and C1.

Therefore, when it is determined that the transfer is completed in units of drive rows (S2060 in FIG. 4: YES), for example, A
When it is determined that the transfer of the drive array is completed, the buffer 8 corresponding to each of the disks A1 to A4 of the A drive array is determined.
The data to be read can be immediately generated based on the data stored in 1a to 84a (S2070 and S2080 in FIG. 4).

As a result, there is no waste in storing the transferred data once and then rearranging the data as data for each disk device 10, and the data is organized simultaneously with the data transfer from the disk device 10. In addition, the time required for data arrangement can be reduced, and the speed of data read processing by parallel processing between drive rows can be increased.

On the other hand, in the process of writing data in parallel to a plurality of drive rows, in the conventional disk array 100, buffers 31 to 34 provided for buses 41 to 44 as shown in FIG. The data to be written transferred from the host device 500 to the cache memory 60 is stored in the order of the write requests.

Accordingly, the 3 connected to each of the buses 41 to 44
In the case where a transfer request for writing is alternately made from one of the disk devices 10, a data operation for selecting and transferring corresponding data from a series of data stored in each of the buses 41 to 44 according to the transfer request is performed. Is required.
Moreover, since such data operation is required for each of the buses 41 to 44, the speed of the data write process by the parallel processing between the drive trains can be increased in the same manner as the data read process depending on the time required for the data operation. It was difficult.

On the other hand, in the disk array 1 of this embodiment, as shown in FIG. 1, the buses 41 to 44 correspond to the three disk devices 10 connected to the buses 41 to 44, respectively. Three buffers 81a-
81c, 82a to 82c, 83a to 83c, 84a to 8
4c. For example, looking at bus 41, 1
Three disks A1, B1,
Each individual buffer 81a, 81b,
81c.

As described with reference to FIG.
The bit operation means 50 distributes the data to be written transferred from the cache memory 60 to the four buffers corresponding to the disk drives 10 in the corresponding drive row (S1050 in FIG. 3) and stores the data (S1050 in FIG. 3). S106
0). For example, data to be written to the A drive row
The data is distributed and stored in the buffers 81a to 84a.

When there is a data transfer request from the disk device 10, by executing the data transfer process described in FIG. 4, the disk device 10 which has requested the transfer is specified as in the read process (FIG. 4). S201 inside
0), and further specifies a buffer corresponding to the disk device 10 that has requested the transfer (S2020 in FIG. 4).
For example, the disk device 10 that has requested the transfer is the disk A1
, And further specifies the individual buffer 81a corresponding to the disk A1.

For example, the buffer 8 corresponding to the disk A1
Since the data to be written from the host device 500 to be written to the disk A1 in which the data to be written from the host device 500 is distributed is previously stored in the disk 1a, the disk 1a and the buffer 8
When 1a is specified, the switch 71 is switched to transfer the corresponding data from the buffer 81a to the disk A1 (S2040 and S2050 in FIG. 4). As a result, the data corresponding to the disk device 10 that has requested the data transfer can be immediately transferred. For example, regarding the bus 41, when the transfer request is made from the disk A1, the data in the buffer 81a is received, when the transfer request is made from the disk B1, the data is stored in the buffer 81b, and the transfer request is made from the disk C1. In this case, the data in the buffer 81c is transferred.

That is, for example, by storing data to be written to the three disks A1, B1, and C1 connected to the bus 41 in the individual buffers 81a to 81c in advance, the three disks A1, B1, and C1 are stored. Even if there is a data transfer request alternately from, or in any order, the corresponding data can be transferred immediately.

This eliminates the waste of selecting and transferring data corresponding to the disk device 10 that has requested the transfer from the data stored for each of the four buses 41 to 44, and reduces the time required for data operation. Shortening can be achieved. As a result, it is possible to realize high-speed data write processing by parallel processing between drive arrays.

In the disk array 1 of the present embodiment, the storage areas of the twelve disk devices 10 are managed as clusters with the same block address for each drive row as described with reference to FIG. Is assigned to each of the drive rows A to C.

As described above, the host device 500 generally stores data by designating continuous clusters such as clusters “3”, “4”, and “5”. At this time, as can be seen from FIG. 2, since continuous clusters are not allocated to the same drive row, when writing a series of data to clusters “3”, “4”, “5”, for example, 3 "," 4 "," 5 "
When a series of data is read from the drive, the time required for the write processing and the read processing can be reduced by the parallel processing between the drive rows. As a result, the disk array 1 of the present embodiment is particularly effective when sequentially reading video data of a movie or the like stored continuously in a cluster.

Further, in the disk array 1 of the present embodiment, as shown by broken lines in FIG.
Can be added to the buffers 81d to 84d. As a result, one disk device can be connected to each of the buses 41 to 44 of the disk array 1, and a new drive row can be added. As a result, the disk array 1 is a scalable device.

Incidentally, the array controller 3 in the disk array 1 of the present embodiment corresponds to the array controller described in claim 5. As described above, the present invention is not limited to such an embodiment at all, and can be implemented in various forms without departing from the gist of the present invention.

For example, the disk array 1 of the above embodiment
Although the configuration in which data is read from the cluster specified by the host device 500 has been described, a configuration in which data is read ahead from a plurality of continuous clusters subsequent to the cluster number specified by the host device 500 is also conceivable. For example, in the case of video data such as a movie as described above, it is expected that data will be stored consecutively in a cluster. In that case, it is particularly effective.

In the disk array 1 of the above embodiment, the same number of buffers as the disk device 10 is provided as an individual storage area corresponding to each disk device 10. Each buffer may be provided, and the storage area of the buffer may be divided into the number of disk devices to be used. However, the configuration in which the number of buffers can be increased according to the number of disk devices 10 as in the above-described embodiment is economical because the number of buffers can be increased or decreased according to the number of drives, and it is economical. This is advantageous in that a sufficient storage area corresponding to 10 can be secured.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a schematic configuration of a disk array according to an embodiment.

FIG. 2 is an explanatory diagram showing management of storage areas of a plurality of disk devices in a disk array.

FIG. 3 is a flowchart illustrating a command issuing process.

FIG. 4 is a flowchart showing a data transfer process.

FIG. 5 is a block diagram showing a schematic configuration of a conventional disk array.

FIG. 6 is an explanatory diagram showing a data transfer situation in a conventional disk array.

FIG. 7 is an explanatory diagram for explaining a problem in a conventional disk array.

[Explanation of symbols]

1,100 Disk array 3,300 Array controller 10 Disk device 21-24 S
CSI controllers 41-44 Bus 50 Bit operation circuit 60 Cache memory 63 PCI bridge 64 PCI bus 71-74 Switch 91-94 DMAC 97 CPU 98 ROM 99 RAM 31-34, 81a, 81b , 81c, 82a, 82
b, 82c, 83a, 83b, 83c, 84a, 84
b, 84c buffer 500 host device

Claims (5)

[Claims] 1. A plurality of drive rows, each having a plurality of buses and being connected to different buses and being operated in parallel by connecting a plurality of storage devices to the respective buses. The storage device is configured to perform access in units of the drive rows by an array controller, and the storage device does not need the bus when performing processing based on a command for at least one of data reading and writing. In a storage device array having a bus occupation / release function for releasing the bus and occupying the bus during a period when the bus is required, the storage device array includes individual storage areas corresponding to the storage devices, respectively. On the basis of a request from the host device, the bus release period is used for each storage device of the drive array to be read or written. Command issuing means for issuing a processing request command, and when a data transfer request is issued from the storage device which is executing processing based on the processing request command, requesting device specifying means for specifying the storage device which has requested the transfer A storage area specifying unit that specifies a storage area corresponding to the storage device specified by the requesting device specifying unit; and a specifying unit that specifies the storage area corresponding to the storage device specified by the requesting device specifying unit when the request from the host device is a data read request. The corresponding data is transferred from the specified storage device to the storage area specified by the storage area specifying unit. On the other hand, when the request from the host device is a data write request, the data is specified by the storage area specifying unit. Transfer executing means for transferring the relevant data from the storage area to the storage device specified by the requesting device specifying means; If the request from the host device is a data read request, and if there is a drive row for which transfer has been completed, the data to be read out is read from the data stored in the storage area corresponding to each storage device in the drive row. And if the request from the host device is a data write request,
When the data to be written is transmitted from the host device, a data operation unit that distributes and stores the data to be written in the storage area corresponding to each storage device in the drive array to be written is provided. Storage device array. 2. The storage device array according to claim 1, wherein a set of unit storage regions at the same position in each of the storage devices constituting the drive array is a management unit storage region from the host device side, and further, A storage device array in which the management unit storage areas are arranged by striping. 3. The storage device array according to claim 2, wherein when there is a data read request from the host device, the data is continuously arranged after the management unit storage area specified by the read request. A storage device array configured to read data from a predetermined number of management unit storage areas. 4. The storage device array according to claim 1, wherein the individual storage areas respectively corresponding to the storage devices are:
A storage device array configured to be able to be added in accordance with the number of connected storage devices. 5. A configuration in which a plurality of storage devices are connected to a plurality of buses, respectively, so that access is made in units of drive rows, which are sets of storage devices connected to different buses and operating in parallel. The array controller has individual storage areas respectively corresponding to the storage devices. Based on a request from a host device, the bus release period is set for each storage device in a drive row to be read or to be written. Command issuing means for issuing a processing request command by using the command, and when a data transfer request is issued from the storage device which is executing processing based on the processing request command, requesting device identification for identifying the storage device requesting the transfer Means, and storage area specifying means for specifying a storage area corresponding to the storage device specified by the requesting device specifying means. When the request from the host device is a data read request, transfer the corresponding data from the storage device specified by the requesting device specifying unit to the storage area specified by the storage area specifying unit, When the request from the host device is a data write request, transfer execution means for transferring the corresponding data from the storage area specified by the storage area specifying means to the storage device specified by the requesting device specifying means; If the request from the host device is a data read request, and if there is a drive row for which transfer has been completed, data to be read is generated from the data stored in the storage area corresponding to each storage device in the drive row. On the other hand, when the request from the host device is a data write request,
When the data to be written is transmitted from the host device, a data operation means is provided for distributing and storing the data to be written in storage areas corresponding to the respective storage devices in the drive array to be written. And array controller.