JP2014052856A - Disk array control device, disk array control method, and program therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a disk array control device in which even when an abnormality occurs at a storage device, smoothness of a response to an input/output request from the outside can be attained, and reduction of input/output throughput of an entire device can be suppressed.SOLUTION: A control device body 100A executes input/output control of data to one logical storage device 30 that is logical and constituted by a plurality of storage units 32a to 32n by means of an external instruction. The control device body 100A includes: main control means 14 for controlling the entire device; storage control means 22 for executing input/output instruction to a plurality of storage devices; storage device control means 22 for executing input/output instruction to the plurality of storage devices; and writing control means 12 that operates when the input/output instruction is a writing request, and executes, via the storage device control means 22, the writing request. There are further provided: record holding means for temporarily storing a writing request which is generated during restoration processing executed after occurrence of an abnormality in any of the storage devices, when an abnormality occurs in any of the storage devices, and delay writing control means 16 for performing delay writing of data associated with the writing request to the one storage device after restoration processing for the one storage device.

Description

  The present invention relates to a disk array control device, a disk array control method, and a program therefor, and in particular, a disk that takes into account a response when an abnormality occurs in the storage device during data input processing for a plurality of storage devices (disk arrays). The present invention relates to an array control device, a disk array control method, and a program thereof.

  In many disk array control devices, a plurality of individual storage devices (storage) are logically combined into one, and this single logical storage device is the control target. In this case, for example, the redundant data (parity data) is generated while dividing the data sent from the host device into a number that is one less than the number of the individual storage devices, and the divided data and redundant data are stored in a plurality of individual memories. The method of storing in each apparatus is adopted.

As a result, even if a failure occurs in one individual storage device, the data can be reproduced from the write data read from the other individual storage device, and the data is recorded in one individual storage device. Can complete data recording at high speed. The same applies to the reading process.
Data can be safely stored by such input / output control, but it is also required to maintain responsiveness to the host device as much as possible during recovery control when an abnormality occurs.

FIG. 9 shows an example of a conventionally known disk array control device.
The disk array control device 200 shown in FIG. 9 performs input / output processing of data sent from the host device 34 to a disk array composed of a plurality of storage devices (disk devices) 30a, 30b,. Storage devices 30a, 30b,. A storage device control unit 222 that individually selects 30n and executes data input / output control; a write control unit 212 that instructs the storage device control unit 222 to perform data input control (write control); Similarly, it comprises a read control means 220 for instructing data output control (read control) and a main control section 214 for controlling the operation of each control means.

  In the data input / output control, the writing control unit 212 for writing and the reading control unit 220 for reading are separately provided, so that switching from the writing operation to the reading operation (or vice versa). The operation can be performed quickly.

As this type of disk array control apparatus 200, other related technologies as shown in Patent Documents 1 to 3 are known.
Among these, Patent Document 1 discloses that the data write time is increased by the seek time for writing to the replacement area of the defective block, and the transfer performance deteriorates. Column), data to be written to the bad block is stored in the non-volatile memory (the upper left column on page 4), and is written from the non-volatile memory to the replacement area during the idle time of the disk array controller (the upper right column on page 4). A technique is disclosed in which the seek to the alternate area is shifted to the idle time of the disk array control device.

  Further, Patent Document 2 aims to guarantee the maximum response time even when a failure occurs in a disk array control device having a redundant configuration (paragraphs 0002 and 0016). When a request is received, the method stores the data in a buffer, notifies the host device of the end (paragraph 0029), and writes write data restored from another disk device after recovery to the recovered disk device (paragraph 0031). It is disclosed.

  Furthermore, Patent Document 3 discloses that all of the RAID (RAID) are used for the purpose of recovery when a loop abnormality induces an overall abnormal state in one disk device in the disk array control device (paragraph 0013). Discloses a method (paragraph 0026) for turning off and on the power of the disk device.

Japanese Patent Laid-Open No. 3-63971 Japanese Patent Laid-Open No. 11-95933 JP 2003-162380 A

  In the related art disclosed in FIG. 9 above, when an abnormality occurs in any of the storage devices 30a to 30n, the storage device control means 222 performs recovery processing for the abnormality. As processing, it is generally known that retry is repeated several times, or a storage device is soft reset or initialized. Also in the case of the above-mentioned Patent Document 1, there is known a method of performing a power reset that turns off the power of the corresponding storage device and turns it on again.

On the other hand, input / output to the storage devices 30a to 30n is suppressed during the execution of the recovery process for the occurrence of an abnormality in each of the storage devices 30a to 30n.
Therefore, the operation command (input / output request) from the read control unit 220 and the write control unit 212 to the storage device control unit 200A is kept waiting until the restoration process is completed. The input / output request from the logical storage control means 21 to the write control means 212 is also waited for its operation, and as a result, the input / output throughput of the entire logical storage device deteriorates. was there.

  Further, as the recovery process takes longer, the response to the input / output request to the entire logical storage device 30 becomes worse. Furthermore, in the recovery method as in the case of Patent Document 1, the power-on operation of the storage device takes longer than the retry, soft reset, initialization, or the like, so that the response becomes worse.

  That is, in the related technique shown in FIG. 9 and each technique disclosed in Patent Documents 1 to 3, if there is another input / output request during the recovery process of turning the power of the storage device (disk device) off and on, Since the response of the input / output request is waited until the storage device is in a normal operation state, there is a disadvantage that the transfer performance is deteriorated.

[Object of invention]
The present invention is capable of smoothly responding to an input / output request even when there is an input / output request from a host device when an abnormality occurs in the individual storage device, and effectively reduces the input / output throughput of the entire logical storage device. It is an object of the present invention to provide a disk array control apparatus, a disk array control method, and a program thereof that can be suppressed.

In order to achieve the above object, a disk array control device according to the present invention performs data input / output control for a logical storage device formed by logically combining a plurality of storage devices based on a command from a host device. A disk array control device having a control device body to be executed,
The control device body is
Main control means for instructing data input / output processing to the logical storage device based on a command from the host device, and input / output processing for the plurality of storage devices that operate based on the command of the main control means Storage device control means for executing, and write control means that operates when the write request is received from the main control means and executes the write request via the storage device control means,
A memory that temporarily holds a write request generated for one storage device during a recovery process of the one storage device involved in the occurrence of an abnormality in any of the plurality of storage devices Providing holding means;
A delay write control means is provided for performing a delayed write of data relating to the temporarily held write request to the one storage device after the recovery process of the one storage device. To do.

In order to achieve the above object, a disk array control method according to the present invention performs data input / output control for a logical storage device formed by logically combining a plurality of storage devices based on a command from a host device. A disk array control device having a control device body to be executed,
A main control unit provided in the control device main body performs data input processing that is performed on the basis of a command from the host device and is performed on the logical storage device. (Write command process)
Based on this command, the storage device control means operates to control input data from the host device to the logical storage device (write control step),
When an abnormality occurs in any of the storage devices that are the targets of the writing process, the storage device control unit performs a recovery process for the one storage device related to the occurrence of the abnormality (recovery processing step), If there is a write request from the host device to the one storage device during the recovery process, this is temporarily stored in the storage holding means equipped in advance (temporary storage processing step),
After the restoration processing of the one storage device, the delayed write control means provided in advance in the main body of the control device stores the data relating to the temporarily held write request in the one storage device. It is characterized in that the data is inserted (delayed writing process).

In order to achieve the above object, a disk array control program according to the present invention performs data input / output control for a logical storage device formed by logically grouping a plurality of storage devices based on a command from a host device. A disk array control device having a control device body to be executed,
A write control processing function that operates based on a command from the host device, executes data input control to the logical storage device, and writes input data from the host device to the logical storage device;
A recovery processing function for recovering one storage device related to the occurrence of an abnormality when an abnormality occurs in any of the storage devices targeted for the writing process;
Temporary storage processing function for temporarily storing this in the record holding means equipped in advance when there is a write request from the host machine during the recovery process,
And a delay write processing function for performing a delay write on the temporarily stored write request to the one storage device after the recovery processing of the one storage device,
Each of these processing functions is realized by a computer provided in the control device main body.

  Since the present invention is configured and functions as described above, according to this, even if an input / output request is received from a host device when an abnormality occurs in the individual storage device, the input / output request is temporarily held and the error is recovered. Since the configuration is such that delayed writing of data related to the input / output request is executed later, a response to the host device can be made smoothly and a decrease in input / output throughput of the entire logical storage device can be effectively suppressed. It is possible to provide an excellent disk array control apparatus, disk array control method, and program thereof that are not included in the related art described above.

It is a block diagram which shows one Embodiment of this invention. It is a block diagram which shows the example in the case of memorize | storing and managing this about the status information at the time of abnormality recovery in embodiment disclosed in FIG. FIG. 3 is a flowchart illustrating an example of a recovery process (power-off / on operation) when an abnormality occurs disclosed in FIG. 2. FIG. FIG. 3 is an explanatory diagram illustrating an operation example of a data input / output process (write control) performed during a recovery process when an abnormality occurs disclosed in FIG. 2. 5 is a flowchart illustrating an example of an operation procedure of write control during the recovery process in FIG. 4. 3 is a flowchart illustrating an example of delayed write control after recovery processing in the embodiment disclosed in FIG. 1. FIG. 8 is a block diagram illustrating an operation example in the case where read control is performed during the recovery process in the embodiment disclosed in FIG. 1. 3 is a flowchart illustrating an example of read control in the embodiment disclosed in FIG. 1. It is a block diagram which shows an example of related technology.

Hereinafter, an embodiment of the present invention will be described with reference to FIGS.
In the present embodiment, the disk array control device 100 performs data input / output control from the higher-level device 34 to the logical storage device 30 that is a logical group of a plurality of storage devices 32a, 32b,. The control device main body 100A is executed based on the command.

  The control device main body 100A includes a main control means 14 for instructing data input / output processing to the logical storage device 30 based on an instruction from the host device 34, and a plurality of storage devices instructed by the main control means 14. 32a, 32b,..., 32n, storage device control means 22 for executing input / output processing, and write control for controlling write requests from the main control means 14 to the plurality of storage devices 32a to 30n. And means 12.

Further, when an abnormality occurs in any of the plurality of storage devices 32a to 32n, the control device main body 100A is generated for the one storage device during the recovery process of the one storage device related to the occurrence of the abnormality. A record holding means (recovery management table) 18 for temporarily holding the write request is provided.
Further, the control device main body 100A reads out the temporarily held data related to the write request from the other storage device and restores it, and after the restoration processing of the one storage device, the control unit main body 100A delays the writing to the one storage device. Delay write control means 16 is provided.

  As a result, even if an abnormality occurs in any of the storage devices 32a to 32n, the command (write request) from the host device 34 can be continued smoothly without stopping the entire device. Input processing (writing processing) can be executed.

Here, the write control unit 12 described above records the write request for the one storage device generated during the recovery process of the one storage device 32a to 32n described above in the storage holding unit 18 as delayed write information. A write request execution notification function 12a is provided for notifying the host device 34 that the write request has been executed.
As a result, it is possible to continuously respond to a write request from the host device 34 generated during the recovery processing of the storage devices 32a to 32n.

  Furthermore, when the above-described control device main body 100A receives a read request for the plurality of storage devices 32a to 32n from the host device 34 via the main control means 14, the control device main body 100A processes the read request accordingly. Read control means 20 is provided.

  The read control means 20 responds to the read request that has occurred in any of the plurality of storage devices 32a to 32n and that occurred during the recovery processing of the storage devices 32a to 32n related to the occurrence of the abnormality. And a read request execution processing function 20a for retrieving data corresponding to the read request from other storage devices other than the storage device being restored and transmitting the data to the host device 34.

  As a result, it is possible to continue to respond to a read command from the host device 34 generated during the recovery process of the storage devices 32a to 32n without interruption.

Further, the above-described storage device control means 22 operates when an abnormality occurs in any of the plurality of storage devices 32a to 32n, and individually selects a power source for each storage device and performs individual power supply for restoration processing for stop control. Power supply control means 10 having a stop control function 10a is provided.
Thereby, the storage devices 32a to 32n can be individually reset, and in many cases, the storage devices 32a, 32b,..., Or 32n related to the occurrence of the abnormality can be restored to a normal state. It has become.

  The storage device control unit 22 described above further performs a recovery process for the storage device related to the occurrence of the abnormality when any abnormality occurs in any of the plurality of storage devices 32a to 32n, and the storage device that is in the recovery process. The state information storage control function 22a for controlling the storage of the state information in the storage holding unit 18 described above, and the write corresponding to the delayed write control unit 16 after recovery from an abnormality occurring in any of the storage devices A write request output function 22b for requesting delayed writing of data related to the write request when requested is provided.

  As a result, the data storage operation for the storage devices 32a to 32n related to the occurrence of the abnormality after recovery from the abnormality is smoothly executed, and the execution process is performed with a reliable response to the host device.

Here, each component mentioned above is demonstrated concretely.
First, the host device 34 is, for example, a host computer, a cache control device of the disk array, a database management system, or the like, and is a device that transmits an input / output request to the disk array control device 100.

  Further, the power control means 10 described above, when an abnormality occurs in any one of the plurality of individual storage devices 32a to 32n, powers off only the individual storage device related to the occurrence of the abnormality and after the power is turned off. Control the power on. In many cases, the error caused by the software can be recovered by turning the power on again.

  When the write control unit 12 receives a write request from the host device 34 after the power is turned off, the write control unit 12 writes the data in the individual storage device 32 other than the individual storage device related to the recovery operation without making it redundant. Then, the main control means 14 notifies the host device 34 of the completion of writing after the writing of the data without redundancy is completed.

  In this case, the non-redundant writing by the write control unit 12 generates parity data, but does not write to the individual storage device in which the abnormality has occurred, but redundantly writes to the logical storage device 30. It is a writing that does not give sex. The main control means 14 notifies the host device 34 of the completion of writing upon completion of writing of data that is not made redundant in the recovery control by turning on / off the power source.

  By having such a configuration, the disk array control apparatus 100 according to the present embodiment does not cause the host device 34 to wait until the individual storage device can be written again in the recovery process by turning off / on the power supply. The transfer performance can be maintained without deteriorating the response to the host device 34 during the restoration process. The same applies to the disk array control method and the disk array control program.

  In addition, after data that is not made redundant is written, in order to reproduce the data, write data of all the individual storage devices other than the individual storage device in which an abnormality has occurred is required. When another new abnormality occurs in the written individual storage device, the data cannot be reproduced. However, in data writing with redundancy, data cannot be reproduced if a failure occurs in two or more individual storage devices at the same time.

  For this reason, in the present embodiment, during the occurrence of an abnormality in one unit, the performance is maintained by completing the writing normally without redundancy, thereby preventing the transfer performance from deteriorating. As a result, it is possible to employ a practically extremely useful recovery process such as power on / off without degrading the performance.

Further, the main control means 14 mentioned above specifically controls the logical storage device 30 and issues an input / output request (write request or read request) issued by the input / output request means 36 of the host device 34. To process. The write control unit 12 processes a write request requested from the main control unit 14.
The read control unit 20 processes a read request requested from the main control unit 14. The storage device control means 22 processes input / output requests from the read control means 20 and the write control means 12 to the individual storage devices 32a to 32n.

Furthermore, the delayed write control means 16 writes the write request received during the recovery process of the individual storage device 32 where the abnormality has occurred after the recovery process (delayed writing).
The storage holding means (recovery management table) 18 is a data table stored in, for example, a non-volatile memory mounted in the disk array control apparatus 100, and includes delayed write management information 18a and recovery processing status information 18b. Remember.

  The delayed write management information 18a records information related to the write request received by the write control means 12 during the recovery process when an abnormality occurs. The recovery process status information 18b is configured to be recorded by the storage device control means 22 during the recovery process when an abnormality occurs.

<About power off / on control>
Here, the restoration process by power control of the individual storage devices 32a, 32b,..., 32n by the storage device control means 22 will be described.
When an abnormality occurs in an individual storage device among the plurality of individual storage devices 32a to 32n, the storage device control means 22 performs a recovery process for the abnormality. As this restoration processing, it is generally known that the input / output request retry is repeated several times, or the individual storage devices 32a to 32n are soft reset or initialized.

  However, since the input / output to the individual storage devices 32a to 32n is normally suppressed during the execution of the recovery processing, the input / output request from the read control means 20 and the write control means 12 is recovered. Wait until processing is complete. That is, a predetermined time for starting is required from when the power is turned on to when the power is turned on.

  Then, the main control means 14 for making an input / output request to the read control means 20 and the write control means 12 is awaited. For this reason, the longer the power-off / on recovery process is, the worse the response of the entire input / output request of the logical storage device 30 is. However, in this embodiment, such inconvenience is surely improved.

  On the other hand, when the individual storage device 32 is turned off / on, no abnormality occurs and there are many cases where the recovery process is successful, and the frequency of replacing the individual storage device 32 as a hardware failure can be reduced. That is, although power off / on as a recovery process is useful for an error to be solved by retrying an input / output request, the response is deteriorated during the recovery process. It will definitely improve as you did.

[Operation of Embodiment]
Next, the operation of the disk array control apparatus 100 according to this embodiment when an abnormality occurs will be described with reference to FIGS.

<When an error occurs>
When an input / output request from the input / output request means 36 of the host device 34 to the logical storage device 30 is generated (FIG. 3: step S1), the main control means 14 receives information (input / output) from the input / output request means 36. Output address, data length, write data in the case of writing), and if it is a read request, an input / output request is sent together with this information to the read control means 20 and to a write control means 12 if it is a write request. Send in (read / write command process).

  The read control unit 20 or the write control unit 12 selects one of the individual storage devices 32a to 32n from the individual storage devices 32a to 32n based on the input / output address address and the data length received from the main control unit 14. select. A plurality of individual storage devices are selected as necessary. In the example shown in FIG. 3, the individual storage device 32b is selected.

  If the individual storage device to be input / output is determined to be, for example, the individual storage device 32b, an input / output request is passed to the storage device control means 22 (FIG. 3: step S2). The storage device control means 22 reads or writes to the individual storage device 32b based on the input / output request passed (read / write control step).

  At this time, when an error occurs in the input / output to / from the individual storage device 32b (FIG. 3: step S3), the storage device control means 22 first checks the individual storage device 32b in which the error has occurred, and then continues to another individual storage device. The power off / on state is confirmed (FIG. 3: steps S4 and S5), and then the recovery process is performed. At the same time, information in the recovery process is temporarily written in the recovery process status information 18b in the record holding means (recovery management table) 18 together with the input / output request (FIG. 3: step S7 / temporary storage process).

  In the restoration process, the power control means 10 is made to function as described above, the power of the individual storage device 32b in which an abnormality has occurred is turned off (FIG. 3: step S8), and then the power is turned on (FIG. 3: step S9). ). Here, the conditional branching in step S5 and step S6 in FIG. 3 will be described later as “power off standby”.

  If no further input request to the individual storage device 32b is generated during the recovery process, after the recovery process is completed and the normal operation state is set (FIG. 3: step S10), the storage device control unit 22 returns the recovery process status information. The information in the recovery process of the individual storage device 32b written in 18b is cleared (FIG. 3: step S11). If the abnormality continues even after the power is turned on, the occurrence of a failure is notified to the host device 34 (FIG. 3: step S12).

  Next, a write request process when an input / output request to the individual storage device 32b is generated during the recovery process (FIG. 3: steps S9 to S10) will be described with reference to FIGS. explain. The read request processing will be described with reference to FIGS.

(Response to write requests during recovery processing)
First, the process of a write request to the individual storage device 32 during the recovery process will be described with reference to FIGS.
First, when an abnormality occurs in the individual storage device 32b, the power control unit 10 shown in FIG. 4 controls the power-off of only the individual storage device 32b and the power-on after the power-off as described above (FIG. 3). : Steps S8 to S9 / Power supply control step).

  When the write control unit 12 receives a write request from the host device 34 after the power is turned off, the write control unit 12 writes the data in the other individual storage devices 32a to 32n other than the individual storage device 32b without making them redundant. At the same time (FIG. 5: Step S27 / write control step), the write completion is given to the main control means 14. Subsequently, the main control means 14 notifies the host device 34 of the end of writing after completion of the writing of the data without redundancy (FIG. 5: step S28 / main control step). As a result, the write request from the host device 34 can be completed without waiting for the completion of the power off / on recovery process, and the transfer performance can be maintained while performing a useful power off / on recovery process. it can.

  On the other hand, prior to the write control, as shown in FIG. 5, when a write request from the input / output request means 36 to the logical storage device 30 is generated (FIG. 5: step S21), the main control means 14 inputs / outputs. The data requested to be written, the data length, and the address are received from the request unit 36 and are passed to the write control unit 12.

  The write control means 12 generates redundant data (parity data) based on the address and data length received from the main control means 14 (FIG. 5: Step S22), and writes the write data including the redundant data respectively. The individual storage device to be written is selected from the individual storage devices 32a to 32n (FIG. 5: Step S23). In this example, the individual storage device 32b and other individual storage devices are selected.

  When the individual storage device and data to be written are specified, the writing control means 12 confirms the recovery processing status information 18b in the record holding means (recovery management table) 18 (FIG. 5: step S24). If there is no individual storage device 32b that is being restored, write data is written to another individual storage device with redundancy (FIG. 5: step S25), and the completion of writing is notified to the host device 34.

  In the example shown in FIGS. 4 and 5, since the individual storage device 32b to be written is being restored (one unit in the above step S24), the write control means 12 delays the record holding means (recovery management table) 18. In the write information management information 18a, the write location and data length of data to be written to the restored individual storage device 32b are recorded, and the necessary data is written without redundancy to the individual storage devices other than the individual storage device 32b. (FIG. 5: Step S27). Then, the host device 34 is notified of the completion of writing (FIG. 5: step S28).

  That is, the write control means 12 records the address and data length of data that could not be written to the individual storage device 32b in the delayed write management information 18a in the record holding means (recovery management table) 18 (FIG. 5: step). S29).

  As described above, in this embodiment, when there is one individual storage device 32 during the recovery process, the writing is completed by writing without redundancy. Thereby, it is possible to execute a recovery process by turning the power off and on without deteriorating the response.

<Writing wait>
In the disk array control apparatus 100 in this embodiment, the write control means 12 has a write standby function 12b, and the main control means 14 has a write hold notification function 14a. If there are one or more individual storage devices other than the individual storage device 32b as described above, the write control means 12 writes until the number of individual storage devices being restored becomes one or less. It is configured to wait for control.

  The write standby function 12b of the write control means 12 shown in FIG. 4 has two or more individual storage devices that are powered off among the plurality of individual storage devices 32a to 32n (FIG. 5: step S24 / 2 or more). Then, the writing without redundancy is waited until the number of the individual power-off individual storage devices becomes one (FIG. 5: step S30). Then, the write hold notification function 14a of the main control unit 14 notifies the host device 34 of the temporary hold during the standby by the write standby function 12b (FIG. 5: step S31).

  As described above, when there are two or more individual storage devices that are powered off, writing without redundancy is waited until the number of the individual storage devices becomes one, and the host device 34 is notified of temporary suspension of writing during this waiting. For an event that has a low probability that two or more individual storage devices will fail at the same time, maintenance is performed by simplifying the overall configuration without preparing a function for preventing the host device 34 from waiting. On the other hand, when an abnormality occurs in one unit, responsiveness can be maintained by writing without redundancy. That is, an optimal function can be provided at a minimum cost.

<Delayed write control>
Further, the disk array control apparatus 100 according to the present embodiment includes a delayed write control means 16 as shown in FIGS.
The delayed write control means 16 operates in response to a request (command) from the storage device control means 22 when the individual storage device 32b is restored by turning on the power, and is written to an individual storage device other than the individual storage device 32b. Is read out based on the record of the delayed write information management information 18a to restore the write data, and the write data is written to the restored individual storage device 32b via the storage device control means 22 (delayed write step). ).

  In this way, the delayed write control means 16 restores the write data having redundancy from the data written without redundancy after the individual storage device 32 is restored by power off / on, and the individual storage device By writing to 32b, an unstable state without redundancy can be resolved immediately after the individual storage device 32b is restored.

  In this case, in this embodiment, the target recording function 16a of the delayed writing control means 16 sets the address and data length of the data to be written to the individual storage device 32b whose power is off, as described above, the delayed writing management information 18a. May be recorded in the record holding means (recovery management table) 18 (FIG. 5: step S29). Then, the target specifying function 16b of the delayed writing control means 16 refers to the delayed writing management information 18a after recovery and specifies data to be restored (FIG. 6: step S42).

  In this way, the address and data length of the delayed writing target are recorded in the record holding means (recovery management table) 18, and the delayed writing is made by referring to the memory holding means (recovery management table) 18 after recovery. The target to be included can be specified.

  More specifically, with respect to this delayed write control, prior to the delayed write control of the recovered data, as shown in FIG. 6, first, the entire apparatus is normally operated after the power is turned on (FIG. 6: When the restoration process is completed in step S10), the storage device control unit 22 requests the delayed write control unit 16 for delayed write control. In response to this, the delayed writing control means 16 confirms the delayed writing management information 18a (FIG. 6: step S41). If there is the delayed writing management information 18a, the delayed writing management information 18a Data for delay writing is specified (FIG. 6: step S42).

  Based on the delayed writing management information 18b, data necessary for generating data to be written to the individual storage device 32b is read from the individual storage device 32 other than the individual storage device 32b to generate write data (FIG. 6: Step S43). Further, the write data is written to the restored individual storage device 32b through the storage device control means 22 (FIG. 6: Step S44), and the information of the delayed write management information 18a is updated every time the writing is completed (FIG. 6: Step S45).

  After step S45 in FIG. 6, the process returns to step S41 in FIG. 6. When the control for all delayed write management information 18a is completed, the delayed write control means 16 notifies the storage device control means 22 of the delayed write completion. To do. Upon receiving the notification, the storage device control means 22 clears the information related to the individual storage device 32b in the recovery process status information 18b.

<Read control>
Next, processing of a read request from the individual storage device 32 during recovery processing will be described with reference to FIGS.

  First, the read control means 20 shown in FIG. 7 reads the data read from the individual storage devices 32a to 32n other than the individual storage devices 32a, 32b,. The data of the read request is restored from and transmitted to the host device 34. That is, the read control means 20 responds to a read request generated during the recovery process of the individual storage device 32a, 32b,..., Or 32n in which an abnormality has occurred, and the individual storage devices 32a to 32a other than the individual storage device being recovered Data corresponding to the read request is reproduced from 32n.

  As shown in FIG. 8, when a read request for the logical storage device 30 is generated from the input / output request means 36 shown in FIG. 7 (FIG. 8: step S51), the main control means 14 reads the data to be read from the input / output request means 36. The data length and address are received and passed to the read control means 20.

  The read control unit 20 selects the individual storage device 32 from which data is read based on the read address and read data length received from the main control unit 14 from the individual storage devices 32a to 32n. In the example shown in FIG. 8, the individual storage device 32b is selected. When the physical disk to be read and the data length are determined, the recovery process status information 18b in the recovery management table 18 is confirmed (FIG. 8: Step S52).

  If there is no individual storage device in the recovery process (FIG. 8: Step S52 / None), it is normal control and data is read with or without redundancy (FIG. 8: Step S53). (FIG. 8: Step S54).

In the example shown in FIG. 7, since the individual storage device 32b to be read is being restored (FIG. 8: one unit in step S52), the data read from the individual storage device 32b is read from the individual storage devices 32 other than the individual storage device 32b. (FIG. 8: Step S55) The data is reproduced (restored), the read control is continued as if it were read from the individual storage device 32b, and the completion of reading is notified (FIG. 8: Step S56).
By this read control, the data transfer performance can be maintained while adopting the power off / on recovery process.

<Reading standby>
In the present embodiment, the read control means 20 includes a read standby function 20b. When there are two or more individual storage devices that are powered off among the plurality of individual storage devices 32a to 32n (FIG. 8: step S52 / 2 or more), this read standby function 20b Reading is waited until there is one (FIG. 8: step S57).
That is, when there are two or more individual storage devices that are powered off, data cannot be restored from redundant write data, and therefore, the reading is waited until there is one individual storage device 32 that is powered off.

  In the example shown in FIGS. 7 and 8, the main control means 14 includes a read suspension notification function 14b. The read hold notification function 14b notifies the host device 34 of a temporary hold during standby by the read standby function 20a described above (FIG. 8: step S58). For an event with a low probability that a failure occurs in two or more individual storage devices 32 at the same time, the entire configuration can be simplified without preparing a function for preventing the host device 34 from waiting. The disk array control apparatus 100 originally does not have redundancy that can cope with the occurrence of two or more simultaneous failures.

  That is, it is assumed in the design that the response deteriorates when an event with a low probability of recovery processing of two units occurs at the same time. For this reason, increasing the probability of successful power-off / on recovery processing due to standby by the read standby function 20b improves the overall transfer performance rather than unnecessarily replacing the individual storage device and causing long-term response deterioration. It is preferable from the viewpoint of maintenance.

  As described above, according to the present embodiment, in the related technology described above, when an abnormality occurs in any of the individual storage devices 32a to 32n, input / output to the corresponding individual storage device is waited for recovery processing. However, in this embodiment, without waiting for the input / output during the above-described recovery processing, the read request reproduces data from the individual storage device other than the individual storage device 32, and the write request stores the write information. By completing the processing once when it is stored in the individual storage device of the disk array control device 100 and performing delayed writing after the recovery processing, the input / output processing is performed, so that it is possible to prevent the throughput from being deteriorated.

<Power off standby>
Here, in addition to the above-described embodiment, as another preferred embodiment, the power control unit 10 may include a power-off standby function 10a. In this case, when there is an individual storage device that is powered off, the power-off standby function 10a waits for power-off of other individual storage devices until the individual storage device is in an operating state after the power is turned on.

  As a result, since the individual storage devices that are turned off are not turned off until the individual storage devices that are turned off are put into operation by turning on the power, the number of individual storage devices that are being restored is always one. Even if a write error occurs in the individual storage device 32, the transfer performance of this read can be maintained by the power-off standby function 10a when the read can be normally handled.

  In FIG. 3, when an input / output error occurs (FIG. 3: step S3), the storage device control means 22 identifies the individual storage device in which the error occurred (FIG. 3: step S4). Then, the power control unit 10 checks whether or not the other individual storage device is powered off (FIG. 3: step S5). For example, the presence / absence of the individual storage device 32 during the recovery process is confirmed with reference to the recovery process status information 18b. When there is an individual storage device 32 that is undergoing recovery processing, the power-off standby function 10a waits for completion of recovery of the individual storage device (FIG. 3: step S6).

  As a result, when the power-off recovery process is always set to 1 or less and a bad block is identified and normal input / output is possible to other blocks, the power off / Recovery processing by turning on can be performed.

  Here, the operation contents (particularly, each operation step in FIGS. 3, 5, 6 and 8) in the operation part of each configuration described above are programmed so as to be executable by a computer, and the above-described disk array. You may comprise so that the computer with which the control apparatus 100 was equipped previously may be made to perform.

  In this case, the programmed program may be recorded on a non-volatile recording medium such as a hard disk, DVD, CD, or flash memory. In this case, this program is read from these recording media by a computer and executed.

  Since this embodiment is configured and functions as described above, according to this, even when an input / output request is received from the host device 34 when an abnormality occurs in the individual storage device, the input / output request is temporarily held. In addition, since the data related to the I / O request is restored to the individual storage device affected by the error after the error is restored, the response to the host device can be made smoothly even when an error occurs in the individual storage device. In addition, it is possible to obtain an excellent effect that is not obtained in the related art described above, that is, it is possible to effectively suppress a decrease in input / output throughput with respect to the entire logical storage device.

  As described above, the present invention has been described with the specific embodiments shown in the drawings. However, the present invention is not limited to the embodiments shown in the drawings, and is known so far as long as the effects of the present invention are exhibited. Other technical elements can be employed.

  About each embodiment mentioned above, it is as follows when the summary of the novel technical content is put together. In addition, although part or all of the said embodiment is put together as follows as a novel technique, this invention is not necessarily limited to this.

[Appendix 1]
A control device main body 100A that executes data input / output control for the logical storage device 30 formed by logically combining a plurality of storage devices 30a, 30b,. A disk array control device 100 comprising:
The control device main body 100A is
Main control means 14 for instructing data input / output processing to the logical storage device 30 based on a command from the host device 34, and the plurality of storage devices 32a to 30n that operate based on the command of the main control means 14 And the storage device control means 22 for executing the input / output processing, and operates when there is the write request from the main control means 14, and executes the write request via the storage device control means 22. Writing control means 12;
When an abnormality occurs in any of the plurality of storage devices 32a to 30n, a write request generated for the one storage device during the recovery process of the one storage device related to the abnormality occurrence is temporarily A memory holding means 18 for holding is provided, and
Provided is a delayed write control means 16 for performing delayed writing of the data related to the temporarily held write request to the one storage device after the recovery processing of the one storage device. A disk array controller.
[Appendix 2] (Write control means)
In the disk array controller according to appendix 1,
When the write request to the one storage device generated during the recovery process of the one storage device is recorded and held in the record holding unit 18 as the delayed write information, the write control unit 12 A disk array control device comprising a write request execution notification function for notifying the host device that the write request has been executed.
[Appendix 3] (Reading control means)
In the disk array controller according to appendix 1,
When a read request for the plurality of storage devices 32a to 32n is sent from the host device via the main control means 14, a read control means 20 is provided for processing the read request,
In response to the read request that occurs during the recovery process of the storage device 32a to which the abnormality occurs in any of the plurality of storage devices and the storage device 32a to which the abnormality has occurred, the read control unit 20 A disk array control device comprising a read request execution processing function for retrieving data corresponding to the read request from another storage device other than the storage device and transmitting the data to the host device.
[Appendix 4] (Individual stop control of disk power supply)
In the disk array controller according to appendix 1,
The storage device control unit 22 operates when an abnormality occurs in any of the plurality of storage devices, and individually selects a power supply of the storage device related to the occurrence of the abnormality and performs individual power control for recovery processing. A disk array controller having a stop control function.
[Appendix 5] (Request for delayed writing)
In the disk array controller according to appendix 4,
When an abnormality occurs in any of the plurality of storage devices, the storage device control unit 22 performs a recovery process for the storage device related to the occurrence of the abnormality, and stores status information of the storage device during the recovery process. State information storage control function for controlling storage in the record holding unit 18, and data related to the write request when there is a corresponding write request to the delayed write control unit 16 after the storage device is restored A disk array control device comprising a write request output function for requesting delayed writing of a disk array.
[Appendix 6] (Method invention / Appendix 1)
A control device main body 100A that executes data input / output control for the logical storage device 30 formed by logically combining a plurality of storage devices 30a, 30b,. In the disk array control apparatus 100 provided,
The main control unit 14 provided in the control device main body 100A performs data input processing that is performed based on a command from the host device 34 and is performed on the logical storage device 30 via the write control means 12. To the storage device control means 22 (write command process),
Based on this command, the storage device control means 22 operates to control input data from the host device 34 to the logical storage device 30 (write control step),
When an abnormality occurs in any of the storage devices 30a to 30n that is the target of the writing process, the storage device control unit 22 performs a recovery process on one storage device related to the occurrence of the abnormality (restoration process). Step), if there is a write request from the host device 34 to the one storage device during the restoration process, this is temporarily stored in the storage holding means 18 provided in advance (temporary storage processing step). ),
After the temporary storage of the data related to the write request after the restoration processing of the one storage device, the delayed writing control means 16 pre-installed in the control device main body 100A stores the data in the one storage device. A disk array control method characterized in that delayed writing is performed (delayed writing step).
[Appendix 7] (Program invention / Appendix 6)
A control device main body 100A that executes data input / output control for the logical storage device 30 formed by logically combining a plurality of storage devices 30a, 30b,. In the disk array control apparatus 100 provided,
A write control processing function that operates based on a command from the host device 34, executes data input control to the logical storage device 30, and writes input data from the host device 34 to the logical storage device 32;
A recovery processing function for recovering one storage device related to the occurrence of an abnormality when an abnormality occurs in any of the storage devices 30a to 30n that is the target of the writing process;
A temporary storage processing function for temporarily storing a write request from the host machine 34 during the restoration process in the record holding means 18 provided in advance;
And a delay write processing function for performing a delay write on the temporarily stored write request to the one storage device after the recovery processing of the one storage device,
A disk array control program characterized in that each of these processing functions is realized by a computer provided in the control device main body 100A.

  The present invention is applicable to a disk array control device using a plurality of individual storage devices 32a to 32n. In particular, in the case of the disk array control apparatus 100 that controls redundant data writing, the present invention can exert its effect remarkably.

DESCRIPTION OF SYMBOLS 10 Power supply control means 10a Individual power supply stop control function 12 Write control means 12a Write request execution notification function 14 Main control means 16 Delayed write control means 18 Record holding means (recovery management table)
18a Delayed write management information 18b Recovery processing status information 20 Read control means 20a Read request execution processing function 22 Storage device control means 22a State information storage control function 22b Write request output function 30 Logical storage devices 32a, 32b,..., 32n Individual Storage device 100 Disk array control device 100A Control device body

Claims (7)

A disk array control device having a control device main body that executes data input / output control on a logical storage device formed by logically combining a plurality of storage devices based on a command from a host device,
The control device body is
Main control means for instructing data input / output processing to the logical storage device based on a command from the host device, and input / output processing for the plurality of storage devices that operate based on the command of the main control means Storage device control means for executing, and write control means that operates when the write request is received from the main control means and executes the write request via the storage device control means,
When an abnormality occurs in any of the plurality of storage devices, a record that temporarily holds a write request generated for the one storage device during the recovery process of the one storage device related to the occurrence of the abnormality Providing holding means;
A delay write control means is provided for performing a delayed write of data relating to the temporarily held write request to the one storage device after the recovery process of the one storage device. Disk array control device. The disk array control device according to claim 1, wherein
The write control means, when a write request to the one storage device generated during the recovery process of the one storage device is recorded and held in the record holding means as the delayed write information, A disk array control device comprising a write request execution notification function for notifying the host device that a write request has been executed. The disk array control device according to claim 1, wherein
Provided with a read control means for processing the read request when a read request for the plurality of storage devices is sent from the host device via the main control means,
In response to the read request that occurs during the recovery process of the storage device in which the abnormality occurs in any of the plurality of storage devices and the storage device involved in the abnormality occurs, the read control unit is not the storage device that is in the recovery process. A disk array control device comprising a read request execution processing function for retrieving data corresponding to the read request from another storage device and transmitting the data to the host device. The disk array control device according to claim 1, wherein
The storage device control means is operated when an abnormality occurs in any of the plurality of storage devices, and individually selects a power supply of the storage device related to the occurrence of the abnormality, and individually controls the power supply for recovery processing. A disk array controller having a stop control function. The disk array control device according to claim 4,
When an abnormality occurs in any of the plurality of storage devices, the storage device control unit performs recovery processing of the storage device related to the occurrence of the abnormality and records and holds status information of the storage device during the recovery processing A state information storage control function for storage control in the means, and delayed writing of data related to the write request when there is a corresponding write request to the delayed write control means after recovery of the storage device A disk array control device having a write request output function for requesting. In a disk array control device having a control device main body that executes data input / output control for a logical storage device formed by logically combining a plurality of storage devices based on a command from a host device,
A main control unit provided in the control device main body performs data input processing that is performed on the basis of a command from the host device and is performed on the logical storage device. (Write command process)
Based on this command, the storage device control means operates to control input data from the host device to the logical storage device (write control step),
When an abnormality occurs in any of the storage devices that are the targets of the writing process, the storage device control unit performs a recovery process for the one storage device related to the occurrence of the abnormality (recovery processing step), If there is a write request from the host device to the one storage device during the recovery process, this is temporarily stored in the pre-equipped record holding means (temporary storage processing step),
After the restoration processing of the one storage device, the delayed write control means provided in advance in the main body of the control device stores the data relating to the temporarily held write request in the one storage device. A disk array control method characterized in that a write operation is performed (delayed writing step). In a disk array control device having a control device main body that executes data input / output control for a logical storage device formed by logically combining a plurality of storage devices based on a command from a host device,
A write control processing function that operates based on a command from the host device, executes data input control to the logical storage device, and writes input data from the host device to the logical storage device;
A recovery processing function for recovering one storage device related to the occurrence of an abnormality when an abnormality occurs in any of the storage devices targeted for the writing process;
Temporary storage processing function for temporarily storing this in the record holding means equipped in advance when there is a write request from the host machine during the recovery process,
And a delay write processing function for performing a delay write on the temporarily stored write request to the one storage device after the recovery processing of the one storage device,
And a disk array control program characterized in that each of these processing functions is realized by a computer provided in the control device main body.

Images