JP3250861B2 - Disk device system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high-speed writing / reading process of large-capacity data in a disk drive system using a disk drive such as a hard disk, an optical disk and a semiconductor disk. It is also suitable for command processing of a disk array composed of a plurality of disk devices.

[0002]

2. Description of the Related Art As an increase in the capacity of a file system and an improvement in transaction performance are required, an array-type disk drive mechanism disclosed in Japanese Patent Application Laid-Open No. 2-236714 discloses a method for processing a large amount of data at high speed. There are systems and methods. This method will be described with reference to FIG.

In FIG. 10, reference numeral 10 denotes Read / Wri.
An upper-level computer that issues a te command, and 11 is an array controller that performs data distribution / set control. 11
Among them, 12 is a higher-level computer I / F section that exchanges commands and data with a higher-level computer, 13 is a task control section that performs task execution management, and 14 is a command processing management that optimizes command multiplex processing. And a command processing unit for reorganizing commands from the host computer 12 into commands of the disk array, and for generating, issuing, and terminating commands to the respective disk devices.
Numerals 20 to 20 denote a disk device I / F section for sending and receiving commands and data to and from the disk, and 21 to 25 for writing data.
This is a disk device that performs reading. Note that the command handled here is a SCSI command, which itself has three elements of a command type, a logical block address (hereinafter referred to as LBA), and a transfer length.

First, a plurality of drive devices 21 to 25 are multiplexed (operated in parallel) by the array controller 11 shown in FIG. The command data transferred from the host computer 10 is subjected to address conversion by the command processing management unit 15 of the array controller 11 so as to be distributed to each disk device in a predetermined unit. Under the control of the task control management unit 13 and the interrupt processing unit 15, each command is sent to each disk via the disk device I / F units 16 to 20 in the order of arrival by the first-in first-out method (hereinafter referred to as FIFO method). Devices 21 and 2
Write 5 In the case of reading command data, the divided command data read from each of the disk devices 21 to 25 is read by the array controller 11 as one.
The data is integrated into one set of data and sent to the host computer 10.

[0005]

The conventional method has the following problems.

That is, in the conventional FIFO method in which commands are processed in the order of arrival, when a host computer executes a Read / Write command on a plurality of disks, the continuous data is included in a plurality of command strings for accessing the continuous data. When commands that access irrelevant data are mixed, a head seek occurs for each mixed command.
There is a problem that the continuous data transfer characteristic is significantly reduced.

If a command uses two disks, disk device 0 and disk device 1,
When the disk device 0 is waiting for access, the disk device 1 is reserved for the command even though the disk device 1 is unused, so that a command arriving later uses the disk device 1. And the multiplexing operation is limited.

An object of the present invention is to provide a disk drive system in which the speed of processing is realized by changing the execution order of commands.

[0009]

In order to achieve the above object, the present invention has a disk device and a disk control device for controlling the disk device, and receives a data read or write command from outside, and In a disk device system that reads or writes data, the disk control device includes a receiving unit that receives the command, a first command queue that stores the command, and a total execution time of the stored command reduced. Execution order control means for changing the execution order of the above commands.

[0010]

The command enters the first command queue, and the execution order of the commands is changed so as to shorten the total execution time of the accumulated commands, thereby achieving higher speed.

For example, the execution order control means rearranges the command group in the first command queue in consideration of the continuous area access possibility, and starts execution from the rearranged command, so that the continuous area access data is transferred. They can be processed together.

Thus, by processing the divided continuous area access data collectively as one command group, it is possible to reduce the seek of the disk head and the like, thereby realizing high speed operation.

Also, in consideration of the possibility of multiple executions of the command group in the first command queue, commands can be executed multiple times by rearranging them and starting execution from the rearranged commands.

Thus, commands can be rearranged so that a plurality of disk devices can execute multiple operations as much as possible.
The operating rate of the disk can be increased, and the average processing time per command can be reduced.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described with reference to FIGS.

The present disk device system (disk array) shown in FIG. 11 receives a command from a host computer 10 that issues a Read / Write command and performs an array controller 11 (which controls data distribution and collection). A disk controller) and disk devices 21 to 25 for writing and reading data. The array controller 11 includes a higher-level computer I / F unit 12 that exchanges commands and data with a higher-level computer, a task control unit 13 that performs task execution management, and a command that optimizes command multiplexing processing. A command processing and management unit 141 for performing address conversion for reorganizing commands from the host computer 10 into commands of the disk array, and generating, issuing, and terminating commands to the respective disk devices; an interrupt processing unit 15; And disk device I / F units 16 to 20 for sending and receiving commands and data to and from a disk. Note that the command handled here is a SCSI command, which itself has three elements of a command type, a logical block address (hereinafter referred to as LBA), and a transfer length.

First, a plurality of drive devices 21 to 25 are multiplexed (operated in parallel) by the array controller 11 shown in FIG. The command data transferred from the host computer 10 is subjected to address conversion by the command processing management unit 15 of the array controller 11 so as to be distributed to each disk device in a predetermined unit. Under the control of the task control management unit 13 and the interrupt processing unit 15, each command is sent to each disk via the disk device I / F units 16 to 20 in the order of arrival by the first-in first-out method (hereinafter referred to as FIFO method). Devices 21 and 2
Write 5 In the case of reading command data, the divided command data read from each of the disk devices 21 to 25 is read by the array controller 11 as one.
The data is integrated into one set of data and sent to the host computer 10.

FIG. 1 is a diagram showing details of the command processing and management unit 141.

The command processing and management unit 141 includes a host computer command receiving unit 30 (receiving means) for receiving a Read / Write command from a host computer, and a Read.
From the LBA and transfer length of the / Write command, an address conversion unit 31 for reorganizing the command sent from the host computer into a command of the array disk, and a Read / Wri
and a command scheduler 32 for rearranging the execution order of the te command.

The command scheduler 32 includes a command waiting queue 33 (second command queue) for arriving commands, a command rearranging queue 34 (first command queue) used for rearranging commands, and a command waiting queue. 33, a queue control unit 35 (queue control unit) for controlling the command rearrangement queue 34, a queue size limiting unit 36 for limiting the size of the queue, and an execution order optimization control unit 37 (execution order control unit). .

The command execution order optimization control unit 37 rearranges commands for accessing continuous areas, and rearranges continuous area access command rearrangement means 38, and rearranges multiple executable commands, and executes multiple execution possibility rearranging means 39.
And

The overall operation of FIG. 1 will be described with reference to the flowchart of FIG.

The host computer command receiving section 30 shown in FIG.
Receives the Read / Write command (200 in FIG. 2), and the address conversion unit 31 performs an address conversion for distributing data to the array disks to the command (202). The command scheduler 32 takes in the commands and sets them in the command waiting queue 33 in the order of arrival (No. 204). The command passes through the command waiting queue 33 and enters the command rearrangement queue 34. By providing a certain condition, the queue size limiting unit 36 stops passing commands through the command waiting queue. Therefore, the number of commands in the command rearrangement queue is limited (No. 206). The conditions include, for example, a method of sending a timing for limiting the queue after a certain time elapses by a timer, and a method of limiting the number of commands arriving at the command rearrangement queue in advance. Further, a command arriving from now on waits in the command waiting queue 33 until the rearranging of the previously arrived command in the command rearranging queue 34 is completed. For commands in the command rearrangement queue, based on the order of arrival (No. 208), first, the continuous area accessibility (No. 210) and then the possibility of multiple execution (No. 212) are checked, and both exist. In such a case, the commands are rearranged by giving priority to the continuous area access possibility, and the commands are rearranged in consideration of the multiplex execution possibility only for commands that do not access the continuous area. If either one exists, the commands are rearranged independently (step 214). Further, the execution is started from the command whose rearrangement has been completed (216). When the command reordering queue becomes empty (218), the command stop queue in the command waiting queue 33 is released, and a new command that has been waiting is put in the command reordering queue 34. Thereafter, the above operations are repeated.

A case where only the possibility of multiple executions is considered will be described as a first embodiment.

FIG. 3 shows the command scheduler 32 of FIG.
Hit. In FIG. 3, reference numeral 33 denotes a command waiting queue in which an incoming command waits, 34 denotes a command rearrangement queue used for rearranging commands, 40 denotes a queue control unit that controls the queues of 33 and 34, and 41 denotes a queue size. A timer for deciding the timing, 42 is an address bit pattern creation unit. 43 is a first bit pattern reading unit for reading the bit pattern of the first command and the command whose sorting has been completed, 44 is a second bit pattern reading unit for reading the bit pattern of a command that has not been sorted, and 45 is 43. , 44
This is a comparator for comparing the bit patterns read by.

The overall operation of FIG. 3 will be described with reference to the flowchart of FIG.

The command scheduler shown in FIG. 3 takes in the read / write command whose address has been converted (flowcharts 300 and 302 in FIG. 4) and places them in the command waiting queue 33 in the order of arrival. At this time, the bit pattern creation unit 42 creates a bit pattern indicating the disk device used by the command based on the LBA unique to the command. Here, the bit pattern indicates the disk device to be used by 1 and the disk not to be used by 0,
11 is a bit sequence corresponding to each of the disk devices in FIG. 10 from the left. For example, when the disk devices No. 0 to 2 are used and the disk devices No. 3 and 4 are not used, the bit pattern is 11100. The bit pattern creation unit 42 holds the created bit pattern in association with the command (304). Therefore, the first bit pattern reading unit 43 reads and inverts the bit pattern of the head command of the queue (308). If there is a 1 in the inverted bit pattern, that is, if there is a vacant disk device, the second bit pattern reading section 44 reads the bit pattern of the remaining commands to be rearranged, and the comparator 45 reads it. The bit pattern is compared with the bit pattern of the first bit pattern reading unit on a bit-by-bit basis (310), and the command having the most matching 1 is selected. Thereafter, the queue control unit 40 rearranges the commands after the first command (312). Further, the execution is started from the rearranged command (314), and when the command rearrangement queue becomes empty (316), the command stop queue 33 is released from the command passing stop, and the new waiting command is replaced with the command. Sorting queue 34
, And the above operation is repeated. bit·
The use of a pattern is a simple and effective means in a rearrangement method for selecting commands so that disk devices to be used do not overlap as much as possible in order to increase the degree of multiplexing of disk accesses.

As a specific example, a command rearrangement operation when ten Read / Write commands as shown in FIG. 3 have arrived at the command rearrangement queue 34 will be described. Here, C # indicates the arrival number of the command, (######
#) Indicates the bit pattern of each command.
First, the first bit pattern reading unit 43 reads the bit pattern 11100 of C1 (using disks 0, 1, and 2) and inverts this information to obtain 00011. Next, the second bit pattern reading unit 44 sets C2
To C10, and the comparator 45 searches for a command with many ones that match or match this inverted bit pattern (all 0s must correspond). In this example, there is no corresponding command, so C2 is the second command. Next, the first bit pattern reading section 43 reads the bit pattern 00101 of C2 and inverts the bit pattern to obtain 11010. Here, as in the case of C1, the second bit pattern reading unit 44 reads the bit patterns from C3 to C10, and the comparator 45 searches for the corresponding command. Then, C9 of 11000 is found, the execution order is changed, and it is arranged next to C2. C2 and C9 are multiplex execution. In order to further determine the possibility of multiplex execution, the queue control unit 40 creates EORs (mask both 0 bits) 00010 of C2 and C9, and the first bit pattern reading unit 43 reads them. Further, the second bit pattern reading unit 42
The bit patterns of C8 and C10 are read from 3 and the comparator 43 searches for the corresponding command. However, since there is no corresponding command, the queue control unit 35 rearranges C3 to the fourth command. Hereinafter, this operation is repeated. FIG. 9A shows the rearranged result.

According to this specific example, each command processing is performed by multiplex operation as much as possible. For example, in the above example, assuming that the command processing time is equal to 1 for all commands,
Before and after the execution order is changed, the total processing time of 10 commands is reduced from 10 when the present invention is not used to 6 by using the present invention, and there is a 40% improvement effect. Therefore,
This multi-execution optimization can increase the I / O throughput of the command.

Further, by providing two queues, the rearrangement of new commands is performed after all the rearranged commands are executed. Therefore, the commands which are used in the conventional case where only one queue is used are used. The sinking (for example, a command in which a low-priority command remains without being executed in the queue forever) can be eliminated.

As a second embodiment, a case in which only the continuous area accessibility is considered will be described.

The command scheduler 532 shown in FIG.
Of the command scheduler 32. In FIG. 5, reference numeral 33 denotes a command waiting queue in which an incoming command waits, 34 denotes a command rearrangement queue used for rearranging commands, 50 denotes a queue control unit that controls the queues of 33 and 34, and 41 denotes a queue size. A timer 51 for determining the timing is a continuous area determination address creating means. 537 is an execution order optimization control unit. 51
Among them, 52 is a first address reading section for reading the address of the command having the lowest command number among the head command and the commands whose sorting has not been completed yet, and 53 is a first address reading section.
A transfer length reading unit for reading a command transfer length included in 2
54 is an adder that calculates the address of a continuous area by adding the transfer length to the head address of the command read by the first address reading unit 52, and 56 is a second address reading unit that reads the addresses of commands that have not been rearranged yet , 57 are comparators for comparing the addresses of 51 and 56.

The overall operation of FIG. 5 will be described with reference to the flowchart of FIG.

The command scheduler shown in FIG. 5 takes in the read / write command whose address has been converted (flowcharts 400 and 402 in FIG. 6) and places them in the command waiting queue 33 in the order of arrival. The command first passes through the command waiting queue and enters the command rearrangement queue 34, and after a predetermined time has elapsed by the timer 41, limits the size of the rearranged queue (404). New commands arriving after this wait in the command waiting queue.
Next, the first address reading unit 52 reads the address of the command at the head of the command rearrangement queue, and the transfer length reading unit 53 reads the transfer length (406).
The transfer length is added to the read address by the adder 54 to calculate the address of the continuous area (408), and the second address reading unit 55 reads the address of the remaining commands to be rearranged, and Compares it with the continuous area determination address creation means 51 (see 41).
0), the matching commands are rearranged after the first command (412). When the command whose execution order is determined is started (step 414) and the command rearrangement queue becomes empty (step 416), the command waiting queue 33 is executed.
Is released, the new command that has been waiting is put in the command rearrangement queue, and the above operation is repeated.

As a specific example, a command rearrangement operation when ten Read / Write commands as shown in FIG. 5 have arrived at the command rearrangement queue 34 will be described. C # indicates the arrival number of the command, (#, #) indicates the address of the command on the left, and the transfer length on the right. First, the first address reading unit 52 in the continuous area determination address creating means 51 reads the address 0 of C1. The transfer length reading unit 53 reads 16 transfer length blocks of C1 (1 block = 512B). Calculates 16 by adding 16 to address 0. Next, the second address reading unit 55 reads the addresses from C2 to C10, and the comparator 56
Search for a command having an address equal to the value 16 calculated in. Then, since C3 is found, the queue control unit 50 rearranges C3 after C1. In order to further determine the possibility of continuous area access, the first address reading unit 52 reads the address 16 of C3, and the transfer length reading unit 53 reads the transfer length 16 block of C3.
When a command is searched by the same operation as described above, C6 is found. At this point, the commands are rearranged in the order of C1, C3, and C6. Thereafter, when the possibility of continuous area access is determined again for C6, C2 is rearranged after C6 because there is no corresponding command. Hereinafter, this operation is repeated. FIG. 9B shows the rearranged result.

According to this specific example, assuming that the number of accesses to one command is 1, before changing the execution order, only C8 and C9 can be accessed in a continuous area, and the number of accesses is 9. After the execution order is changed, C1 and C
Since 3, C6 and C5, C7, C10 are also added, the number of accesses is 5. In general, by serializing a sufficiently large number of commands, the entire disk array can be accessed, and the head seek time can be reduced. There is also an effect of preventing an increase in head seek time when other commands are mixed in long data.

Further, when accessing long data,
The prefetch buffer of the disk device can be used. That is, in the case of a command having short data with discontinuous addresses, it is necessary to rewrite the contents of the prefetch buffer for each command. The processing time is increased by increasing the number of times of rewriting.

The present embodiment is effective not only for a disk array but also for a single disk device.

As a third embodiment, a case where only the possibility of multiple executions is considered will be described.

FIG. 7 shows the command scheduler 32 of FIG.
Hit. In FIG. 7, reference numerals 61 to 70 denote command arrival queues corresponding to combinations of k disk devices (queues prepared by the number of all combinations for selecting k disk devices from n disk devices), and 71 to 80. Is a flag indicating that the command arrival queue has sent a command to the command execution queue, and 81 is a command execution queue waiting for execution of a command whose sorting has been completed. Reference numeral 60 denotes a queue control unit for controlling the command arrival queues 61 to 70 and flags 71 to 80; 42, an address bit pattern creation unit; and 82, a command rearrangement table reference unit.

Next, the overall operation of FIG. 7 will be described with reference to the flowchart of FIG.

The command scheduler 732 shown in FIG.
Imports a Read / Write command whose address has been converted (flowcharts 500 and 502 in FIG. 8), and the bit pattern creation unit 42 creates a bit pattern indicating the disk used by the command based on the LBA unique to the command. Next, the queue control unit 60 controls each of the command arrival queues 61 to 70 corresponding to the bit pattern.
(504, 506). When all the commands at the head of the command arrival queue are collected, the queue control unit 60 selects the command with the smallest command number (508), and transfers that command to the command execution queue 81.
Is set, and the flag of the queue which sent the command among the flags 71 to 80 is set (510). The queue corresponding to the queue that sent the command is referred to by the command rearranging table reference means 82 (512). If there is no corresponding command, the command with the next lower command number is sent to the command execution queue 81, and the flag 71 ~ 80
The flag of the queue that sent the command is set (510). If there is a corresponding command, the command is sent to the command execution queue 81, and the flag of the queue that sent the command among the flags 71 to 80 is set (fig. 5).
14). The above operation is performed until all the flags 70 to 80 of the command arrival queues 61 to 70 are set (516).
When the rearranged commands arrive in the command execution queue 81, the command is executed (518), and when the command execution queue 81 becomes empty (520), all the flags 71 to 80 of the command arrival queues 61 to 70 are executed. Is canceled (522). Then, again, the command arrival queues 61 to 61
Among the commands 70, the command with the youngest command number is sent to the command execution queue 81, and the operation up to now is repeated.

As a specific example, a plurality of read / write accesses to two disk devices as shown in FIG.
A command rearrangement operation when a command has arrived at the command arrival queues 61 to 70 will be described. Here, C # represents the arrival number of the command. 732 is a command scheduler, 61 to 70 are command waiting queues in which incoming commands wait, 60 is a queue control unit that controls the queues of 61 to 70, and 737 is an execution order optimization control unit. First, with all commands at the head of the command arrival queues 61 to 70, the queue control unit 60 assigns C1 with the smallest command number to the command execution queue 81.
And set the flag 71. Since C1 uses the 0th and 1st disk devices, this command and a command that can be executed in a multiplexed manner are sorted into a command rearranging table reference means 8
2 (see the table in FIG. 8). Then, since the commands using the third and fourth disk devices are appropriate, the queue control unit 60 sends C7 to the command execution queue and sets the flag 80. Next, the queue control unit 60 sends out C2 having the smallest command number from the command arrival queues for which no flag has yet been set to the command execution queue, and sets the flag 76. Similar to the operation of C7, when searching for a command that can be executed in multiplex with C2, C10 is found, and the queue control unit 60 sends C10 to the command execution queue 81 and sets a flag 79. Hereinafter, this operation is repeated. Fig. 9 shows the result of rearrangement.
It is shown in (c).

According to this example, since all command processing is performed in a multiplex operation, assuming that the command processing times are all equal to 1, the processing time of 10 commands is changed before the execution order is changed. Later it will be 5. In general, before the execution order is changed, the command processing time changes in a range of 5 to 10 (from completely executing multiple commands to not executing multiple commands at all) depending on the arrangement of commands. Will always be 5, so a maximum of 50
% Improvement effect.

A modification of the second embodiment will be described as a fourth embodiment.

There is a method of adding a means for regarding a command group put together in a continuous area to the command scheduler 32 of FIG. 5 as one command. This is because, as is clear from FIG.
By reconstructing the commands into one command, the total number of processing commands is reduced, and the prefetch buffer of the disk device can be used effectively for the combined commands regardless of the data size. In the case of no data, there is often no data in the prefetch buffer, so that a mishit is likely to occur), and furthermore, it is possible to realize a high-speed Write process while avoiding the rotation waiting time of the disk device.

In particular, in a disk array, Writ
Since the read processing associated with the parity update peculiar to the e-time can be reduced (for example, when data addresses are not continuous, for example, when data is different, the parity disk may be different. Reading is required), and the writing process can be sped up.

From the above description, it is possible to rearrange the commands in consideration of both the continuous area accessibility and the multiple execution possibility, and to use the first embodiment and the second embodiment in the configuration shown in FIG. . Further, the use of the third embodiment and the second embodiment has the same effect as the case where the first embodiment and the second embodiment are used. Further, by using the fourth embodiment instead of the second embodiment, a greater effect can be obtained. As an effect, before the execution order is changed, by changing the execution order even when all commands have Read / Write access, the command execution time is reduced to commands that can be accessed in a continuous area and commands that can be executed multiple times. Be shortened.

[0049]

According to the present invention, the total execution time of a command can be reduced.

Further, by executing commands that can be accessed in a continuous area collectively, there is an effect that the head seek and rotation wait of the disk drive can be reduced.

Also, by executing commands in a multiplex manner, the operation rate of the disk device can be increased, and the I / O rate can be up to 50% of the conventional one.
There is an effect of improving the / O throughput.

Furthermore, since the size of the queue is limited, commands are rearranged in the queue, and the commands are executed in the rearranged order, the effect of preventing sinking of commands in which commands that are not executed remain in the queue forever is prevented. There is.

[Brief description of the drawings]

FIG. 1 is a block diagram of a command processing and management unit according to the present invention.

FIG. 2 is a flowchart according to the present invention.

FIG. 3 is a block diagram showing a first embodiment of a command scheduler.

FIG. 4 is a flowchart according to the present invention.

FIG. 5 is a block diagram showing a second embodiment of the command scheduler.

FIG. 6 is a flowchart according to the present invention.

FIG. 7 is a block diagram showing a third embodiment of the command scheduler.

FIG. 8 is a flowchart according to the present invention.

FIG. 9 is an explanatory diagram showing a command rearrangement result according to each embodiment.

FIG. 10 is a block diagram showing a basic configuration of a disk array.

FIG. 11 is a block diagram showing a basic configuration of a disk array.

[Explanation of symbols]

10 ... upper computer, 11 ... array controller, 12 ... upper computer I / F, 13 ...
Task control unit, 14 command processing and management unit,
15: Interrupt processing unit, 16 to 20: Disk device I / F, 21 to 25: Disk device, 2
9 ... execution order optimization command processing and management unit, 30
······ Higher-level computer command receiving unit 31 ······ Address conversion unit 32 ····· Command scheduler
33 command queue, 34 command rearrangement queue 35 queue controller 36
... Queue size limiting means 37 ... Execution order optimization control unit 38 ... Continuous area access possibility rearranging means 39 ... Means of change,
40: queue control unit, 41: timer, 4
2 ····· Bit pattern creation unit, 43 ····· First bit pattern reading unit, 44 ····· Second bit
Pattern reading unit, 45 ... comparator, 50 ...
··· Queue control unit, 51 ···· Continuous area determination address creation means, 52 ····· First address reading unit, 53 ·
······ Transfer length reading unit, 54 ····· Adder, 55 ·
········ Second address reading unit, 56 ····· Comparator,
60... Queue controller, 61 to 70... Command arrival queue, 71 to 80... Flag, 81.
.. Command execution queue, 82... Command reordering table reference means.

──────────────────────────────────────────────────続 き Continued on the front page (72) Takashi Oeda Takashi Oka, Microelectronics Equipment Development Laboratory, Hitachi, Ltd. 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa-ken 292, Hitachi, Ltd. Microelectronics Equipment Development Laboratory (56) References JP-A-3-183067 (JP, A) JP-A-3-260823 (JP, A) JP-A-55-91049 (JP, A) JP-A-3-220620 (JP, A) JP-A-3-231368 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06F 3/06 G11B 20/10 G11B 20 / 12

Claims (4)

(57) [Claims] 1. A disk array comprising a plurality of disk devices.
And a disk control device for controlling the plurality of disk devices.
Read or write data from outside
Read the above data or
In a disk device system that performs writing, the disk control device includes a receiving unit that receives the command, a command queue that stores the command, and a disk queue of a plurality of disk devices that form the disk array.
So that multiple executions that operate each
The execution order of commands stored in the command queue
Execution order control means for controlling the execution of a plurality of disk devices constituting the disk array.
Commands to operate each of them in parallel
Means for issuing to each of the
Disk device system. 2. The disk drive system according to claim 1,
The execution order control means indicates a configuration of a disk device included in the disk array.
Multiple execution from the command queue based on the
By searching for possible commands
Disk device system characterized by changing the execution order
Tem. 3. The disk drive system according to claim 2,
In addition, the execution order control means further includes a plurality of disk devices constituting the disk array.
For each corresponding data set, the
Disk device characterized by repeating command search
system. 4. The disk drive system according to claim 1,
The command queue has a plurality of predetermined disks.
The execution order control means uses the command received by the receiving means in response to the command.
The command based on the information indicating the disk device to be used.
Distributed to Dokyu was definite command to be executed in the command queue
In some cases, this command and
Identify the command queue to be separated and identify the identified command
By taking out the command in the queue, the command
Disk unit system characterized by changing the execution order of
Stem.