JP2000172570A - Data input / output method and reconstruction method in information recording device - Google Patents

Abstract

(57) [Problem] To provide an information recording device capable of guaranteeing the real-time property of real-time data, or executing the data reconstruction process while guaranteeing the real-time property. A control unit (33) converts an input access request into a read request or a write request for a disk array (2) having a predetermined priority, and creates the read request or write request. It is queued in the queue 36 according to a predetermined priority. Request selection unit 37
Selects a request to be processed by the disk array 2 from read requests or write requests queued in the request queue 36 according to a predetermined priority. The disk array 2 processes the read request or the write request selected by the request selection unit 37.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information recording apparatus, and more particularly, to a redundant array composed of a recording medium for recording redundant data and a redundant array according to an access request transmitted from a host device. The present invention relates to an information recording device including an array controller that controls an array.

[0002]

2. Description of the Related Art In recent years, an information recording apparatus capable of recording moving picture information and / or audio information has been desired. Video information and / or
Alternatively, since the audio information needs to be output without interruption from the information recording device, real-time processing is required for inputting and outputting data to and from the information recording device. In order to fulfill this request, a SCSI (Small Computer System Interface) may be used as an input / output interface of the information recording apparatus. Hereinafter, in this specification, data that requires real-time properties such as moving image information or audio information is referred to as “real-time data”. Data that does not require real-time properties is referred to as “non-real-time data”. The non-real-time data includes title information and total time information of moving image information and / or audio information.

[0003] In recent years, an information recording apparatus having a recording medium arranged in a redundant array has attracted attention. The redundant array of recording media often includes a plurality of recording media, and redundantly records the redundant data. Also,
In the recording medium in the redundant array, error detection redundant information is recorded on at least one recording medium. More specifically, the information recording apparatus includes one or more recording media arranged in a redundant array, and an array controller that controls input and output of data to and from the redundant array. The array controller accesses the redundant array to write and read the redundant data, thereby improving the reliability of the information recording device and improving the data input / output speed. Note that a recording medium that is a disk is called a disk array. For disk arrays, see A Case for Redundant Array of Inexpensive Di
sks (RAID) by David A. Patterson, Garth Gibson, and
Randy. Katz University of California Berkeley (De
cember 1987).

[0004] A parity group is formed on a recording medium in a redundant array. Even if one of the recording media belonging to the parity group fails completely or partially, the data on the failed recording medium is transferred to another recording medium belonging to the same parity group by using a known error correction technique. (Reconstruction processing). However, this rebuilding process cannot be executed if another recording medium belonging to the same parity group fails when a certain recording medium has failed, so it needs to be executed as quickly as possible.

An example of such a reconstruction process is disclosed in Japanese Patent Laid-Open Publication No.
No. 127839 ". The disk controller of the information recording device monitors the operation state of the disk array, and when a rebuilding process is necessary, variable speed playback,
Either idle time reproduction or opportunistic reproduction reconstruction processing is selected and executed. Variable speed playback is performed after interleaving with the I / O (access) operation of the disk array. In addition, the idle time reproduction is executed using the idle time of the disk array. Further, opportunistic reproduction is performed every time an error is detected in the track being accessed.

[0006]

In the redundant array of the information recording device, real-time data and non-real-time data are recorded together. Therefore, the information recording device,
An access request for accessing real-time data and an access request for accessing non-real-time data are transmitted from outside.

However, in the information recording apparatus and the input / output interface such as SCSI described in the above-mentioned publications and literatures, it is not considered that the two types of access requests are separately processed. Therefore, when traffic on the transmission path increases due to a non-real-time data access request, the input / output of real-time data is adversely affected. That is, there is a problem that the information recording apparatus cannot guarantee the real-time property of the real-time data.

In the information recording apparatus having the above-described redundant array, the array controller executes the rebuilding process for the failed disk while processing the data processing request transmitted from the host. Such a conventional rebuilding process has been executed as a whole for each parity group. However, in RAID4 or RAID5, since each disk operates independently,
Even if one disk is idle, another disk may be under load. for that reason,
The conventional method of controlling the parity groups as a unit has a problem that the data reconstruction process cannot be executed efficiently.

Therefore, an object of the present invention is to provide an information recording device capable of guaranteeing the real-time property of real-time data. Another object of the present invention is to provide an information recording apparatus capable of executing data reconstruction processing while guaranteeing real-time performance.

[0010]

Means for Solving the Problems and Effects of the Invention In order to achieve the above-mentioned object, the present invention has the following characteristic structure, and thereby has the following effects. First
The present invention relates to a data input / output method used for an information recording apparatus including a redundant array comprising a recording medium for recording redundant data, and an array controller for controlling the redundant array in accordance with an access request transmitted from a host device. Wherein the array controller creates a read request or a write request for a redundant array having a predetermined priority based on the input access request, and the array controller internally stores the created read request or write request. Queued according to a predetermined priority, and the array controller selects, from the read requests or write requests queued in the queue, the one to be processed by the redundant array according to the predetermined priority, The redundant array is responsible for the read requests or To process the write request.

According to the first aspect, the array controller converts the input access request into a read request or a write request having a predetermined priority. In the redundant array, a read request or a write request selected by the array controller according to the priority is processed. for that reason,
An information recording device including a redundant array in which redundant data is recorded generates a read request or a write request having a relatively high priority for an access request requiring real-time processing, and does not require real-time processing. An access request having a relatively low priority can be created. Thus, the information recording device can distinguish the access request of the host device according to the necessity of the real-time processing. as a result,
An access request that requires real-time processing is processed in the information recording apparatus without being adversely affected by an access request that does not require real-time processing.

A second invention is dependent on the first invention,
The array controller includes a queue corresponding to a priority therein, and the created read request or write request is queued in a queue corresponding to a predetermined priority.

According to the second aspect of the present invention, since the queues are provided in accordance with the priorities, it is possible to distinguish the access request of the host device according to the necessity of the real-time processing, Various processes in the apparatus are performed efficiently.

A third invention is dependent on the first invention,
The array controller internally includes a queue corresponding to a predetermined priority for each recording medium, and the array controller has a predetermined priority based on the input access request and a read request for each recording medium. Or, a write request is created, and the array controller assigns the read request or the write request created for each recording medium according to a predetermined priority.
It is characterized by queuing in a queue of a corresponding recording medium.

According to the third aspect, the queue is provided for each recording medium and in accordance with the priority, so that the access of the host device is performed for each recording medium according to the necessity of the real-time processing. A request can be selected, and various processes in the information recording device are performed more efficiently.

A fourth invention is dependent on the first invention,
The predetermined priority is set based on the real-time processing of the processing in the redundant array.

According to the fourth aspect, the predetermined priority is set based on the real-time property, so that an access request that requires real-time processing is adversely affected by an access request that does not require the real-time property. Instead, it is processed in the information recording device.

A fifth invention is according to the first invention,
When the input / output interface with the host device conforms to the SCSI, the predetermined priority is set in advance in the LUN or LBA field of the access request.

According to the fifth aspect, since the predetermined priority is preset in the access request, the host device determines the priority of the read request or the write request, that is, the priority of the read request or the write request. It becomes possible to notify the information recording device whether or not the level of priority is required.

A sixth invention is an information recording apparatus which includes a redundant array comprising a recording medium for recording redundant data and controls the redundant array in accordance with an access request transmitted from a host device. A control unit configured to generate a read request or a write request for a redundant array having a predetermined priority based on the input access request; and a read request or a write request generated by the control unit is configured to have a predetermined priority. A request queue that is queued in a queue included therein, and a selecting unit that selects, from a read request or a write request queued in the queue, one to be processed by the redundant array according to a predetermined priority. The redundant array processes the read request or the write request selected by the selection unit.

According to the sixth aspect, the input access request is converted into a read request or a write request having a predetermined priority. In the redundant array, a read request or a write request selected by the selection unit according to the priority is processed. Therefore, an information recording device including a redundant array in which redundant data is recorded creates a read request or a write request having a relatively high priority for an access request requiring real-time processing, and the real-time processing is performed. Unnecessary access requests can be created by those having a relatively low priority. by this,
The information recording device, according to the need for real-time processing,
The access request from the host device can be distinguished. As a result, an access request that requires real-time processing is processed in the information recording device without being adversely affected by an access request that does not require real-time processing.

A seventh invention is dependent on the sixth invention,
The request queue includes a queue corresponding to the priority therein, and the read request or the write request created by the control unit includes:
It is characterized by being queued in a queue corresponding to a predetermined priority.

According to the seventh aspect, since the queues are provided in correspondence with the priorities, it is possible to distinguish the access request of the host device according to the necessity of the real-time processing, and to record the information. Various processes in the apparatus are performed efficiently.

An eighth invention is dependent on the sixth invention,
The request queue includes a queue corresponding to a predetermined priority for each recording medium therein, and the control unit has a predetermined priority based on the input access request, the read request for each recording medium or A write request is created, and a read request or a write request created for each recording medium by the control unit is queued in a queue of the corresponding recording medium according to a predetermined priority.

According to the eighth aspect, the queue is provided for each recording medium and in accordance with the priority, so that the access of the host device is performed for each recording medium according to the necessity of the real-time processing. A request can be selected, and various processes in the information recording device are performed more efficiently.

According to a ninth invention, an information recording apparatus includes a redundant array comprising a recording medium for recording redundant data and an array controller for controlling the redundant array in accordance with an access request transmitted from a host device. A data reconstruction method for reproducing and reconstructing data recorded on a failed recording medium that is a failed recording medium in the redundant array, wherein the array controller has a predetermined priority. A read request or a write request for the redundant array required for the construction process is created, and the array controller queues the created read request or write request in a queue therein according to a predetermined priority. , Created from those queued in the queue according to the given priority The read or write request selected, the redundant array processes the selected read or write request, and the array controller reassembles the data based on the result of the read or write request being processed by the redundant array. Execute the construction process.

According to the ninth aspect, the array controller creates a read request or a write request related to the data rebuilding process. The created read request and write request have a predetermined priority. In the redundant array, a read request or a write request selected by the array controller according to the priority is processed. Therefore, if the information recording device that executes the data restructuring process assigns a relatively low priority to the read request and the write request related to the data restructuring process, these requests adversely affect other real-time processes. It is processed without being given. On the other hand, if the information recording device assigns a relatively high priority to the read request and the write request, these requests are processed with priority, so that it is possible to guarantee the completion time of the data reconstruction process. it can.

A tenth invention is according to the ninth invention, wherein the array controller includes therein a queue corresponding to a predetermined priority for each recording medium, and the array controller performs a data reconstruction process. A read request or a write request for each recording medium having a predetermined priority is created, and the array controller generates a read request or a write request for each recording medium according to a predetermined priority. It is characterized by queuing in a medium queue.

According to the tenth aspect, the queue is provided for each recording medium and in accordance with the priority, and further, the array controller makes a read request or a write for each recording medium having a predetermined priority. Since the request is created, the access request of the host device can be distinguished for each recording medium according to the necessity of the real-time processing, and various processes in the information recording device are performed more efficiently.

An eleventh invention is according to the ninth invention, wherein the read request and the write request created by the array controller are assigned a relatively low priority.

According to the eleventh aspect, since the read request and the write request have relatively low priorities, they are processed without adversely affecting other real-time processes.

A twelfth aspect is according to the ninth aspect, wherein a read request and a write request created by the array controller are given a relatively high priority.

According to the twelfth aspect, since the read request and the write request have relatively high priority, they are processed preferentially. This guarantees the completion time of the data reconstruction process.

The thirteenth invention includes a redundant array comprising a recording medium for recording redundant data, and an array controller for controlling the redundant array according to an access request transmitted from a host device, and further comprising the redundant array. A data input / output method used in an information recording apparatus that reproduces data recorded on a failed recording medium, which is a failed recording medium in an array, and reconstructs the spare recording medium. When an access request for the data to be transmitted is transmitted from the host device to the information recording device, the array controller reads the reproduction data necessary for reproducing the data recorded on the failure recording medium from the redundant array, The array controller uses the reproduction data read from the redundant array to Performing an operation to reproduce the data recorded on the faulty recording medium, the array controller creates a write request with a predetermined priority to write the reproduced data to a spare recording medium, The controller queues the created write request in a queue internally provided according to a predetermined priority, and the array controller determines that the created write request is to be processed by the redundant array according to the predetermined priority. , Select from the queue, the redundant array processes the write request selected by the array controller and writes the reproduced data to a spare recording medium.
A relatively low priority is assigned.

[0035] According to the thirteenth aspect, the array controller, when the access request for the data to be reconstructed on the spare recording medium is transmitted from the host device,
The data requested to be accessed is reproduced and written to a spare recording medium. Therefore, when the information recording apparatus subsequently executes the data reconstruction process, it is not necessary to reproduce the data requested to be accessed. As a result, the time required for the data reconstruction process is reduced.

[0036]

DESCRIPTION OF THE PREFERRED EMBODIMENTS In an embodiment of the present invention, RA
ID is the Redundant Array of Inexpensive (Independe
nt) Disks. SCSI stands for Small Co.
Abbreviation for mputer System Interface.

FIG. 1 is a block diagram showing a configuration of an information recording apparatus 1 according to one embodiment of the present invention. In FIG. 1, the information recording apparatus 1 has a RAID3 and RAID4 architecture, and includes a disk array 2 and an array controller 3. The disk array 2 includes first to fifth
Disk devices 21 to 25 (only disks 21 and 25 are shown), and each of the disk devices 21 to 25 is formed into a redundant array. First to fourth disk devices 21 to 21
Reference numeral 24 stores redundant real-time data and / or non-real-time data. Fifth disk device 2
Reference numeral 5 stores parity, that is, error detection redundant information (hereinafter, referred to as a parity block). In the present embodiment, the first to fifth disk devices 21 to 25 form one parity group “A”.

The array controller 3 includes a host interface unit 31, a required rank determining unit 32, and a control unit 33.
And the first to fifth non-priority queues 341 to 345 and the first
A request queue 36 having fifth to fifth priority queues 351 to 355, a request selection unit 37, a disk interface unit 38, a parity operation unit 39, and a buffer management unit 310.
And a table storage unit 311.

The host interface unit 31 is an input / output interface with the host device 4. In the present embodiment, the input / output interface conforms to SCSI. Here, the host device 4 will be described. The host device 4
An access request is made to the information recording device 1. Here, when the input / output interface conforms to the SCSI, the host device 4
Requests access as follows. First, the host device 4 acquires the right to use the SCSI bus by arbitration, and then specifies the information recording device 1 to which an access request is requested by selection. Next, the host device 4
By transmitting Identify (see FIG. 2A), which is one of the SCSI messages, the LUN (Logical Unit No.) of the information recording apparatus 1 to be accessed is specified. Next, the host device 4 transmits a Simple Queue Tag (see FIG. 2B), which is one of the SCSI messages, to issue a plurality of access requests to the information recording device 1. Next, the host device 4
LB to be transmitted from the information recording device 1 by transmitting Read 10 (see FIG. 3A) as a CSI command
A (Logical Block Address)
And data length, and Write10 (FIG. 3)
(See (b)) to specify the LBA and the data length to be written to the information recording device 1. Further, when the host device 4 has transmitted Write 10, the host device 4 transmits data to be written to the information recording device 1. The access request is made as described above.

The host interface unit 31 has a series of SCSI messages transmitted from the host device 4,
After receiving the SCSI command and the data, they are combined into one and output to the request rank determining unit 32 as a processing request. FIG. 4 shows the format of this processing request. In FIG. 4, the processing request includes a command identification, an identification number, and an LU.
N, control information, LBA, data length, and data. These information elements are created as follows.

As the command type, a number corresponding to the operation code of Read 10 or Write 10 is set. In this embodiment, from the viewpoint of simplification of description, when Read 10 is transmitted, information indicating that the access request is for reading (hereinafter, referred to as “R”) is set as the command type. You. Also, W
When write10 is transmitted, information indicating that the access request is for writing (hereinafter, referred to as “W”) is set as the command type.

As the identification number, the above Simple Q
cue tag specified by ueue Tag (Fig. 2 (b)
(See Reference). As a LUN,
A number corresponding to the LUN specified by the Identify is set. The designated LUN number indicates the priority of the access request specified by the host device 4 and information indicating the parity group to be accessed by the access request. Here, the information recording device 1 includes:
LUN numbers are assigned in advance so as to be different from each other for each parity group and as information indicating the priority. As a result, the host device 4 determines the priority (that is,
It is possible to notify the information recording apparatus 1 of how much real-time property is required) and the parity group to be accessed. In the information recording device 1 of the present embodiment,
It is assumed that “non-priority” and “priority” are set as priorities, and “0” and “1” are assigned to the parity group “A” as non-priority and priority information.

As the control information, cache control information such as DPO and FUA specified by the above Read 10 or Write 10 is set. Read as LBA
10 or a value corresponding to the LBA specified by Write 10 is set. As the data length, a value corresponding to the transfer data length specified by Read 10 or Write 10 is set. As the data, when the host interface unit 31 receives Write 10, the Write
The data transmitted in connection with te10 is set. This data is not set when the host interface unit 31 receives Read10.

There are four types of processing requests generated by the host interface unit 31 when classified based on the above-described command type and LUN. That is, first, the first processing request including “R” and “0” is transmitted to each of the disk devices 21 to 21.
This is created when the host device 4 requests the information recording device 1 to read non-real-time data in the data group 25 (parity group “A”). In addition, the second including “R” and “1”
Is created when the host device 4 requests the information recording device 1 to read the real-time data in the parity group “A”. A third access request including “W” and “0” is created when the host device 4 requests the information recording device 1 to write non-real-time data to the parity group “A”. Further, a fourth access request including “W” and “1” is created when the host device 4 requests the information recording device 1 to write real-time data to the parity group “A”.

The request rank discriminating section 32 is configured to execute the LUN included in the processing request output from the host interface section 31.
Then, information on the priority of the processing request and the parity group is extracted, and the priority of the processing request and the parity group are determined. After that, the request rank determination unit 32 outputs the input access request and the information of the priority and the parity group determined by itself to the control unit 33.

The control unit 33 controls reading and writing according to the input priority and parity group information (see FIGS. 5, 7 and 8), and reconstructs data as required (FIG. 11). To FIG. 14).
The request queue 36 has first to fifth disk devices 21 to
For each 25, a set of two queues is provided. More specifically, the first non-priority queue 341 and the first priority queue 351 are allocated to the first disk 21.
Hereinafter, similarly, the fifth non-priority queue 345 and the fifth
Is assigned to the fifth disk 25.

The request selecting section 37 selects and removes a request to be executed by the disk array 2 from the request queue 36 and outputs the request to the disk interface section 38. The disk interface unit 38 is an input / output interface between the array controller 3 and the disk array 2. The parity calculation unit 39 performs a parity calculation. The buffer management unit 310 manages an internal buffer memory for temporarily storing data. Table storage unit 311
Stores a table indicating the state of the LBA of each of the disk devices 21 to 25.

The information recording apparatus 1 as described above will be described first with reference to the flowchart of FIG.
The operation when the host device 4 requests reading of non-real-time data in the parity group “A” will be described.

Upon receiving the access request (a series of SCSI messages and SCSI commands) transmitted by the host device 4 (step S51), the host interface unit 31 of the information recording device 1 creates the above-described first processing request. Is output to the request rank determination unit 32. The request rank determining unit 32 determines the priority based on the LUN of the input processing request (step S52), and determines the parity group for which access is currently requested. According to the above assumption, the request rank determining unit 32 determines that the LUN
Is “0”, the priority is “non-priority”
Is determined as a parity group and “A” is determined. After this determination, the request rank determination unit 32 controls the input first processing request (see FIG. 4) and the determination result (currently, information of priority “non-priority” and parity group “A”). Output to the unit 33.

If the priority information included in the input discrimination result is "non-priority", the control unit 33 proceeds to step S53.
Execute The control unit 33 replaces the information of the LUN in the input first access request with the information indicating the priority and the information indicating the parity group, and transmits the first access request to the disk array 2 (more specifically, In general,
(For each disk drive 21-24). Further, the control unit 33 controls the parity group “A” by the number of disk devices that store the non-real-time data requested by the first access request.
Create a first read request. Each first read request is:
It has the format shown in FIG. 6, and is different from the processing request of FIG. 4 in that LUN information is changed to information indicating a priority and a parity group. Next, the control unit 3
3 is a first non-priority queue 341 to 341 corresponding to the first to fourth disk units 21 to 24 according to the information of the priority “non-priority” and the parity group “A”.
The created first read requests are queued one by one in the fourth non-priority queue 344 (FIG. 3: step S5).
3).

When any one of the disk devices in the disk array 2 completes the process (read or write), the request selection unit 37 executes the process of step S54. Specifically, the request selecting unit 37 determines that the request (third read request described later) created by the control unit 33 is queued in the priority queue corresponding to the disk device for which processing has been completed. Removes the request from this priority queue,
The execution of the request is instructed to this disk device through the disk interface unit 38. Now, assuming that the processing of the first disk device 21 has been completed, the request selection unit 37
When a request is queued in the first priority queue 351, the request is removed and the first disk device 21
To execute the process.

When the priority queue is empty and a request is queued in the same set of non-priority queues, the request selection unit 37 removes the request from the non-priority queue and performs the same processing as described above. Make instructions. Now, assuming that the processing of the first disk device 21 has been completed, the request selection unit 37
Indicates that if the first priority queue 351 is empty and a request is queued in the first non-priority queue 341, the request is executed. Further, the request selection unit 3
7 if both the priority queue and the non-priority queue are empty,
Wait for another disk device to complete the process. The request selecting unit 37 executes step S54 as described above.

When a request is input through the disk interface unit 38, each of the disk devices 21 to 25 executes the processing as requested (step S55). Therefore, the first read request queued in each of the non-priority queues 341 to 344 is sent to each of the priority queues 3 to 24 by the disk devices 21 to 24 executing this step S55.
It is processed in a non-priority manner as compared with those queued in 51-355. Therefore, each of the disk devices 21 to 2
The non-real-time data in 4 is read without adversely affecting the processing (reading or writing) of the real-time data. Hereinafter, a case where the reading according to the first reading request is completed will be described. Each disk device 2
When the reading is completed normally, the non-real-time data and a signal indicating that the reading has been completed normally (hereinafter, referred to as “reading completed”) are output to the disk interface unit 38. Non-real-time data
The data is stored in the buffer area in the buffer management unit 310 through the disk interface unit 38. Also, “read complete” is transmitted through the disk interface unit 38
It is transmitted to the control unit 33.

On the other hand, each of the disk devices 21 to 24 generates a signal (hereinafter, referred to as “read failure”) indicating that the reading has not been completed normally due to a failure or the like. This “read failure” is transmitted to the control unit 33 through the disk interface unit 38.

The controller 33 controls each of the disk devices 21 to 24
Determines whether the reading of data (currently, non-real-time data) has been completed normally (step S5).
6). When the controller 33 receives the "read completed" of each of the disk devices 21 to 25, it checks whether or not the read non-real-time data is arranged in the buffer area (step S57). When the non-real-time data is available, the control unit 33 instructs the buffer management unit 310 to transfer the non-real-time data in the buffer area to the host device 4 through the host interface unit 31.
It is sent to (Step S 5 8). After this transmission,
The buffer area storing the non-real-time data is released.

On the other hand, when the control unit 33 receives “read failure” from at least one disk device of the parity group “A” in step S56, the control unit 33 determines that reading from all the disks 21 to 25 has not been completed normally. Then, the process proceeds to step S59 to execute a process at the time of abnormal reading. Hereinafter, step S59 will be described with reference to FIG. 7 which is a flowchart showing a detailed processing procedure.

The control unit 33 provides data necessary for reproducing the non-real-time data that has not been read (hereinafter, referred to as data).
A second read request for reading (referred to as reproduction data) is newly created (step S71). Control unit 33
Reconfirms the priority information (currently “non-priority”) input in step S53 (step S72). When the priority is “non-priority”, the control unit 33 queues the created second read request in the non-priority queues 341 to 345 corresponding to the disks 21 to 25 of the parity group “A” ( Step S73).

Thereafter, in the information recording device 1, the same operations as those in steps S54, S55 and S57 of FIG. 5 are executed (steps S74, S75 and S76). Control unit 3
In step S76, when it is determined in step S76 that the above-described reproduction data is present in the buffer area, the instruction to the parity operation unit 39 is performed. The non-real-time data that has not been read is reproduced by this parity operation, and then stored in the buffer area of the buffer management unit 310 (step S77). As a result, the process of step S59 ends, and thereafter, in the information recording device 1, the above-described steps S57 and S57 of FIG.
58 is executed.

Next, referring to the flowchart of FIG.
An operation when the host device 4 requests reading of real-time data in the parity group “A” will be described.

The host interface unit 31 executes the process of step S 51, creates the above-described second access request based on the received SCSI message and SCSI command, and outputs it to the request rank determining unit 32. The request rank determination unit 32 executes the process of step S52 to determine the priority “priority” and the parity group “A” because the LUN information of the second access request input this time is “1”. Determine. Thereafter, the request rank determining unit 32 determines whether or not the second access request has been input (see FIG. 4).
And the determination result (currently, information of priority “non-priority” and parity group “A”) is output to the control unit 33.

If the input priority information is "priority", control unit 33 executes step S510. The control unit 33 replaces the LUN information in the input second access request with the information indicating the input priority and the information indicating the parity group, and converts the second access request into
This is converted into a third read request for the disk array 2 (more specifically, for each of the disk devices 21 to 24). Further, the control unit 33 controls the parity group “A”
, The third read requests are created for the number of disk devices that store the real-time data requested by the second access request. Each third read request has the same format as the first read request (see FIG. 6). Next, the control unit 33 determines the first priority queue 351 to the fourth priority queue corresponding to the first to fourth disk devices 21 to 24 according to the information of the priority “priority” and the parity group “A”. The created first read requests are queued one by one in the queue 354 (FIG. 3;
Step S510).

The request selecting section 37 determines whether the above-mentioned step S5 has been performed.
4 is executed. Thereby, each priority queue 35
As is clear from the above, the third read requests in 1 to 354 are preferentially processed by the respective disk devices 21 to 24 executing the above-described step S55. As a result, the real-time data in each of the disk devices 21 to 24 is read without impairing the real-time property. Hereinafter, since the processing of steps S56 to S59 is as described above, the description thereof is omitted, but the information recording apparatus 1
Is performed on real-time data instead of non-real-time data.

Next, referring to the flowchart of FIG.
An operation when the host device 4 requests writing of non-real-time data in the parity group “A” will be described. In this case, the host device 4 also transmits non-real-time data to be written in the disk array 2.

The host interface unit 31 provides an access request (a series of SCSI messages,
When the SCSI command and the non-real-time data are received (step S81), the above-described third processing request is created and output to the request rank determination unit 32, and the received non-real-time data is stored in the buffer area of the buffer management unit 310. Store. The request rank determination unit 32 determines the priority of the input access request (Step S
82), the parity group for which access is currently requested is determined. According to the above assumption, the required rank determining unit 32
Since the value of the LUN is “0”, “A” is determined as the priority with “non-priority” as the parity group. After this determination, the request rank determination unit 32 sends the input access request (see FIG. 4) and the determination result (currently, information of priority “non-priority” and parity group “A”) to the control unit 33. Output.

When the input priority information is “non-priority”, the control unit 33 first sets data (hereinafter, referred to as “data”) necessary for updating the parity block of the fifth disk device 25.
It is determined whether or not to read update data) (step S83). When the control unit 33 determines that the update data is not read, the process directly proceeds to step S87 described below. On the other hand, if the control unit 33 determines to read the update data,
(See FIG. 4). Next, the control unit 3
No. 3 indicates that the priority of the third processing request is “non-priority”, so the first disk device 21 to the fourth disk device 2
The created fourth read request is queued in the first non-priority queue 341 to the fourth non-priority queue 345 corresponding to No. 4 (FIG. 8; step S84).

The request selecting section 37 and each of the disk devices 21 to 21
24 executes the processing as described in steps S54 and S55 of FIG. As a result, each non-priority queue 341
344 are processed by the respective disk devices 21 to 24, although not prioritized, whereby the update data is read.

When the reading process is normally completed, each of the disk devices 21 to 24 outputs the read update data and "read completed" which is a signal indicating that the reading is completed. The update data is stored in the buffer area of the buffer management unit 310 via the disk interface unit 38. In addition, “read complete” indicates that the control unit 3
3 is sent.

The control unit 33 determines whether or not each of the disks 21 to 25 has completed the data (currently update data) read processing based on the “read completed”. If they are aligned within the area (step S85), the parity calculation unit 39 is instructed to execute the parity calculation. The parity calculation unit 39 executes a parity calculation in response to the instruction, and obtains information of a new parity block corresponding to the current non-real-time data write processing. The control unit 33 generates a new parity block based on the information of the new parity block obtained from the parity operation unit 39. However, the control unit 33
Generates a new parity block based on the transmitted non-real-time data when steps S84 and S85 are not executed.

Next, the control unit 33 creates a first write request for writing a new parity block to the fifth disk device 25 with reference to the information of the new parity block. The control unit 33 then sends the host device 4
Reconfirm the priority of the third processing request output from.
When the priority is “non-priority”, the control unit 33 creates the first non-priority queue in the fifth non-priority queue 345 corresponding to the disk device 25 storing the parity block in the parity group “A”. Is queued (step S86).

The control unit 33 replaces the LUN information in the input third processing request with the input information indicating the priority and the parity group, and transmits the third processing request to the disk array 2. (More specifically, to the second write request of each of the disk devices 21 to 24). Further, the control unit 33 creates second write requests for the number of disk devices capable of recording the requested non-real-time data in the parity group “A”. Note that each second write request has the same format as the first read request (see FIG. 6). next,
The control unit 33 controls the first disk devices 21 to 21 according to the information of the priority “priority” and the parity group “A”.
The created second write request is queued in the first non-priority queue 341 to the fourth non-priority queue 344 corresponding to the fourth disk device 24 (FIG. 8; step S8).
7).

The request selecting section 37 performs the processing in step S54 of FIG.
The processing as described in is executed. As a result, the fifth
The first write request queued in the non-priority queue 345 is processed by the fifth disk device 25, though not in priority. As a result, a new parity block is written to the fifth disk device 25. Further, the first non-priority queue 341 to the fourth non-priority queue 3
Each second write request in 44 is also a non-priority, but the first disk device 21 to the fourth disk device 24
Processed by As a result, the buffer management unit 310
The non-real-time data in the buffer area of
3 and the first disk devices 21 to 21
The data is written to the fourth disk device 24.

When the above-described writing is completed, each of the disk devices 21 to 25 outputs a signal indicating that the writing is completed (hereinafter, referred to as “writing completed”) to the disk interface unit 38. “Writing completed” is transmitted to the control unit 33 through the disk interface unit 38.

The control unit 33 controls each of the disk devices 21 to 25
Is received (step S88), it is determined that the writing of the non-real-time data is completed. Further, the control unit 33 notifies the host device 4 via the host interface unit 31 that the writing of the non-real-time data has been completed (step S89).

Next, referring to the flowchart of FIG.
The operation when the host device 4 requests writing of real-time data in the parity group “A” will be described. In this case, the host device 4 also transmits real-time data to be written in the disk array 2 together.

Upon receiving the access request (a series of SCSI messages, SCSI commands, and real-time data), the host interface unit 31 (step S8)
1) The above-described fourth processing request (see FIG. 4) is created and output to the request rank determination unit 32, and the received real-time data is stored in the buffer area of the buffer management unit 310. Since the LUN value of the fourth processing request is “1”, the request rank determining unit 32 determines the priority “priority” and the parity group “A”, and then determines the fourth processing request and the priority Degree "priority" and parity group "A"
Is output to the control unit 33.

When the input priority information is "priority", the control unit 33 executes steps S810 to S814 similar to steps S83 to S87 described above. Therefore, the following description will focus on the differences. I do.

When the control unit 33 receives the priority information “priority” and the parity group information “A”, and determines that the update data is not to be read, the control unit 33 proceeds directly to step S813 to be described later. If it is determined that the update data is to be read, a fifth read request (see FIG. 6) for reading the update data is created. Since the fifth read request has the priority information of “priority”, the first priority queue 3 corresponding to the disk devices 21 to 24 on which the update data is recorded in the parity group “A” is sent.
It is queued in the 51st to fourth priority queues 354 (FIG. 8; step S811).

The request selection unit 37 and each of the disk devices 21 to 21
24 executes the processing as described in steps S54 and S55 of FIG. As a result, each priority queue 351-
The fifth read request in 354 corresponds to each of the disk devices 21 to
24, the processing is preferentially performed. Each disk device 21
When the read process is completed normally, the read update data and the above-mentioned “read completed” are transmitted to the buffer management unit 310 through the disk interface unit 38.
And outputs it to the control unit 33.

The control unit 33 receives the input “read complete”
It is determined whether or not each of the reading processes has been completed normally on the basis of. If the result of this determination is that update data is available in the buffer management unit 310 (step S81
2) As described above, the parity calculation unit 39 performs the parity calculation, and the control unit 33 generates a new parity block. However, when Steps S811 and S812 are not performed, the control unit 33 generates a new parity block based on real-time data.

Next, after creating a third write request for writing a new parity block to the disk device 25, the control unit 33 reconfirms the priority. Control unit 3
If the priority is “priority” as a result of the reconfirmation, the third write request is queued in the fifth priority queue 355 (step S813).

Next, the control unit 33 transmits the input fourth processing request to the disk array 2 (more specifically, the disk devices 21 to 24) in the same manner as in the case of the third processing request. 4th write request (for FIG. 6)
Convert to The fourth write request is also sent to the first disk device 21 to the fourth disk device 24 corresponding to the first disk device 21.
Are queued in the first to fourth priority queues 351 to 354 (FIG. 8; step S814).

The request selecting section 37 performs the processing in step S54 of FIG.
The processing as described in is executed. As a result, the fifth
The first write request in the priority queue 355 is preferentially processed by the fifth disk device 25, whereby a new parity block is written to the fifth disk device 25. Also, the second write requests in the first priority queue 351 to the fourth priority queue 354 are also preferentially processed by the first disk device 21 to the fourth disk device 24, thereby real-time data Is written to the first disk device 21 to the fourth disk device 24 after being made redundant by the control unit 33.

Each of the disk devices 21 to 25 outputs "writing completed" after completing its own writing process. Each output “writing completion” is transmitted to the control unit 33 through the disk interface unit 38. When the control unit 33 receives the “write completed” of each of the disk devices 21 to 25 (step S88), it determines that the writing of the real-time data requested by the host device 4 has been completed, and through the host interface unit 31. Then, the host device 4 is notified that the writing is completed (step S89).

As described above, the host device 4 sends an access request including information indicating priority and the like to the information recording device 1.
Send to The array controller 3 creates a request (read request or write request; see FIG. 6) for each disk device based on the input access request, and creates a request in a predetermined queue according to the determined priority. Queue the request. Therefore, in the disk array 2,
Requests with high priority are processed preferentially. As a result, a high-priority access request such as moving image information and / or audio information that requires real-time properties and a low-priority access request that does not require real-time properties such as title information of moving image information are hosted. Even when the information is transmitted from the device 4 to the information recording device 1, the processing related to the high-priority access request is not affected by that of the low-priority access request. Even low processing can be processed efficiently.

Next, the data reconstructing process of the information recording apparatus 1 will be described. In the following description, a failed disk means a disk in which recorded data has failed. Also, the data reconstruction process
When a failed disk occurs in the disk array 2, a parity operation is performed using data of another disk belonging to the same parity group to reproduce data of the failed disk, and the reproduced data is recorded on another disk. It is to fix it. In addition, the information recording device 1 according to the present embodiment
Executes the reconstruction processing of two types of data. One is
This is a first data reconstructing process that does not affect the real-time process performed by the information recording device 1, and the other is a preset disk in order to guarantee a time when the data reconstructing process is completed. Is a second data restructuring process that preferentially uses a part of the band.

Also, in the data restructuring process, FIG.
The table storage unit 311 shown in FIG. As shown in FIG. 9, the table storage unit 311 stores first to fifth disk management tables 3111 to 3115 for each of the first to fifth disk devices 21 to 25. First to fifth
The disk management tables 3111 to 3115 of the
To the statuses of all LBAs in the fifth to fifth disk devices 21 to 25 are set. This status is managed using the size of the data block or parity block as a unit. For example, each column of the first disk management table 3111 (one hatched portion indicates one column) has a corresponding LB
The status of A is written.

As shown in FIG. 10, the status types include a "normal state" indicating that the corresponding LBA has not failed, and a "failure" indicating that the corresponding LBA has failed (see FIG. 10). 10, not shown) and the corresponding LB
“Rebuilding required state” indicating that A needs to undergo data restructuring processing, and “Rebuilding” indicating that the corresponding LBA is undergoing data restructuring processing. There is.

First, the disk interface unit 38
When detecting that any one of the disk devices 21 to 25 has become a failed disk device, the controller 33 is notified to that effect. Here, the detection of the failed disk device is performed by a failure notification from each of the disk devices 21 to 25 and a timeout of a response from each of the disk devices 21 to 25.

Upon receiving the notification that the failed disk device has been detected, the control unit 33 accesses the disk management table for the failed disk device in the table storage unit 311. Next, the control unit 33 updates the disk management table being accessed, and sets the status of the currently failed LBA to “failed”. For example, when all the LBAs of the failed disk device have failed, the statuses of all the LBAs are set to “failed” in the corresponding disk management table.

The first data rebuilding process in the case where the first disk device 21 has become the failed disk 21 and all the internal LBAs have failed will be described with reference to FIG.
And a flowchart of FIG. FIG.
FIG. 1 is a flowchart showing the overall processing procedure of the first data restructuring process.

The control unit 33 separates the failed disk device 21 from the parity group "A", incorporates a spare disk device (not shown in FIG. 1) into the same group, and furthermore, a disk management table for the spare disk device. (Not shown in FIG. 9) is created in the table storage unit 311. In the created new disk management table (hereinafter, referred to as a new disk management table), the status for all LBAs is set to “rebuild required”. Further, the control unit 33 allocates a pair of queues assigned to the separated failed disk 21 to a spare disk device.

Next, the control unit 33 checks the first LBA in the new disk management table (FIG. 11; step S111), and the status of the LBA is "rebuild required".
Is satisfied (step S112), the LBA is processed. Next, the control unit 33 controls the buffer management unit 3
The used amount of the buffer area in 10 is equal to or less than a predetermined number “M”, and the number of requests (described later) related to reconstruction queued in each of the non-priority queues 341 to 345 is equal to or less than a predetermined number “N”. It is determined whether or not the condition is satisfied (step S113).

This step S113 is necessary in order to suppress a large number of requests related to rebuilding from occurring at once. The number of requests generated is suppressed for the following two reasons. First, if the number of requests generated is large, there is a possibility that an access request from the host device 4 having the same priority as a request related to reconstruction is not processed forever. For example, if the number of requests related to the currently occurring reconstruction is always less than N, it is guaranteed that the access request from the host device 4 will be processed at the Nth at the latest. The predetermined number “N” is determined based on how much an access request from the host device 4 having the same priority as the request related to the rebuilding needs to be processed during the rebuilding process.

Second, if the number of requests generated is large,
In some cases, a memory (not shown) of the array controller 3 runs short. More specifically, when a request related to reconstruction occurs, a memory for storing information of the request and a memory for storing data when a write process is involved are required. Therefore, if the array controller 3 generates a request related to the reconstruction at once, there is a possibility that the internal memory becomes insufficient. Further, when the internal memory is insufficient, the information recording device 1 cannot receive a request from the host device 4. For example, when the array controller 2 uses a maximum of M buffer areas when the access request from the host device 4 is received at the maximum, the array controller 2 generates a request related to rebuilding, and the remaining buffer area becomes “M”. You have to stop when This predetermined number “M” is determined based on the maximum buffer area number used when the maximum request number is received from the host device 4.

The control unit 33 waits until the condition of step S113 is satisfied, and then executes a rebuilding process for the LBA to be processed (step S114). here,
If the condition of step S113 is satisfied even after the start of the new rebuilding process, the rebuilding process for the second LBA is started. Similarly, the rebuilding process continues to be activated until the condition of step S113 is not satisfied. Hereinafter, reference will be made to FIG. 12 which is a flowchart showing the details of step S114.

First, the control unit 33 updates the status of the processing target LBA from “rebuild required” to “rebuilding” in the new disk management table (step S12).
1) To read data (hereinafter, referred to as reproduction data) necessary to reproduce data to be recorded in the LBA to be processed (currently, data recorded in the failed disk device 21) by parity calculation. A sixth read request (see FIG. 6) is created. Next, the control unit 33
The created sixth read request is queued in the corresponding non-priority queue in the request queue 36 (step S122). Here, at this time, the failed disk device 2
Since 1 is the first disk device 21, the sixth read request is not queued in the first non-priority queue 341, but is queued in the second to fifth non-priority queues 342 to 345.

The request selection section 37 determines whether the above-described step S3
4 (step S123), and the respective disk devices (currently, the second disk device 22 to the fifth disk device 25) perform the requested processing as in step S55 of FIG. Is executed (Step S12)
4). As a result, the second to fifth non-priority queues 342 to 3
The sixth read request in 45 is processed by the second to fifth disk devices 22 to 25, though it is not prioritized. When the reading process is completed in each of the disk devices 22 to 25, the disk array 2 sends the reproduction data and a signal indicating that the reading is completed (hereinafter, referred to as "reading completed") to the disk interface unit 38. Output to The reproduction data is stored in the buffer area in the buffer management unit 310, like the non-real-time data.
Also, “read complete” is transmitted to the control unit 33.

The control unit 33 controls each of the disk devices 22 to 25
It is checked whether or not the reproduction data from is stored in the buffer area (step S125). Control unit 33
Instructs the parity operation unit 39 to execute the parity operation when the reproduction data is available. As a result, the data to be recorded in the LBA to be processed (the data recorded in the failed disk device) is reproduced and then stored in the buffer area of the buffer management unit 310 (step S1).
26).

Next, the control unit 33 creates a fifth write request for writing the data currently stored in the buffer area to the corresponding LBA in the spare disk device, and allocates the fifth write request to the spare disk device. The created fifth write request is queued in the non-priority queue 351 (step S127).

The request selecting section 37 performs the processing in step S54 of FIG.
(Step S128). As a result, the fifth write request in the non-priority queue 351 is processed by the non-priority but spare disk device 21 (step S129). As a result, data in the buffer area is written to the disk device 21. When the writing is completed, the spare disk device creates a "writing completed" indicating that.
“Write completed” is transmitted to the control unit 33 through the disk interface unit 38.

When the control unit 33 receives the "write completed" from the spare disk device (step S1210),
The table management unit 311 updates the status of the processing target LBA currently set to “rebuilding” to “normal state” (step S1211). As a result, the process of step S114 ends, and thereafter, the control unit 33 determines whether or not step S114 has been executed for all the LBAs in the spare disk device in the new disk management table with the status "rebuild required". Is determined by checking whether or not is set (step S115). If step S114 has not been executed for all LBAs, the control unit 33 checks the status of the next LBA (step S116), and
Steps S112 to S116 as described above are executed until step S114 is executed for A.

According to the above-described first data reconstructing process, the requests for reconstructing the data (the sixth read request and the fifth write request) are queued in the non-priority queue. The recording device is a second and fourth high priority
The data reconstruction process can be executed without adversely affecting the processing of the processing request. Further, the first data restructuring process is executed efficiently because the queuing of each request and the execution control are performed for each disk device.

The second data reconstruction process in the case where the first disk device 21 has become the failed disk 21 and all the internal LBAs have failed will be described with reference to FIG.
And a flowchart of FIG. FIG.
FIG. 3 is a flowchart illustrating an overall processing procedure of the second data restructuring processing. 13 is different from FIG. 11 only in that step S113 is changed to step S131. Therefore, in FIG.
Steps corresponding to one step are given the same step numbers, and their descriptions are simplified.

First, as in the case of the first data rebuilding process, the failed disk unit 21 is replaced with a spare disk unit (not shown in FIG. 1), and a new disk unit for the spare disk unit is replaced. A disk management table (not shown in FIG. 9) is created.

Next, the controller 33 executes steps S111 and S112 to search for an LBA to be processed. Next, the control unit 33 determines whether or not a predetermined time “T” has elapsed since the previous execution of Step S114 (Step S131).

By the way, each of the disk devices 21 to 25
Since the bandwidth is finite, if many requests related to reconstruction are processed in a short time, access requests from the host device 4 cannot be processed at all. In this step S131, the speed of the reconstruction process is set in advance in order to control the adverse effect of the request related to the reconstruction on the processing of the access request. The array controller 3 controls the rebuilding process to be executed at the set speed.
For example, if the number of LBAs requiring rebuilding is X, and the speed of the above-described rebuilding process is a speed at which rebuilding of Z LBAs is performed in Y minutes, the number of rebuilds in X / (Z / Y) minutes The construction process ends. Further, the control unit 33 only needs to generate one request related to the reconstruction every Y / Z. That is, in this example, T is chosen such that Z requests per rebuild are generated per Y minutes. As described above, T is determined based on the set speed of the reconstruction process.

When the predetermined time “T” has elapsed, the control unit 33 executes a rebuilding process for the LBA to be processed (step S114). Hereinafter, this step S1
Reference is made to FIG. FIG. 14 is different from FIG.
The only difference is that steps 122 and S127 are replaced with steps S141 and S142. Therefore, in FIG. 14, steps corresponding to those in FIG. 12 are denoted by the same step numbers, and description thereof will be simplified.

The control unit 33 first executes step S121 to update the status of the LBA to be processed to "rebuilding", and, as in the case of the first data rebuilding process, reproduces the data for reproduction. A seventh read request (see FIG. 6) for reading the data is created. Next, the control unit 33 queues the created seventh read request in the corresponding priority queue in the request queue 36 (Step S14).
1). Here, at the present time, the failed disk device 21
Since it is the first disk device 21, the seventh read request is not queued in the first priority queue 351 and the second read request is
To the fifth priority queues 342 to 345.

When the request selecting section 37 executes step S123, step S124 is executed in each disk device (currently, the second disk device 22 to the fifth disk device 25). Therefore, in the reconstructing process of the second data, the seventh read request is preferentially processed by the request selecting unit 37 and each disk device. When the reading is completed in each of the disk devices 22 to 25, the disk array 2 sends the read reproduction data and “read completed” to the disk interface unit 3.
8 is output. The output reproduction data is stored in the buffer area of the buffer management unit 310 like the real-time data and the like. The “reading completion” indicates that the control unit 3
3 is sent.

The control unit 33 determines in steps S105 and S1
Step 06 is executed to reproduce data to be recorded in the LBA to be processed (data recorded in the failed disk device), and store the reproduced data in the buffer area of the buffer management unit 310. Next, the control unit 33 creates a sixth write request and places the created sixth write request in the priority queue 351 assigned to the spare disk device, as in the case of the first data construction process. Is queued (step S122).

The sixth write request in the priority queue 351 is preferentially processed by the spare disk device 21 by the request selecting unit 37 executing step S128 (step S129). As a result, after the data in the buffer area is recorded in the disk device 21, “writing completion” similar to the case of the first data reconstructing process is performed via the disk interface unit 38 by the control unit. 33.

The control unit 33 determines in steps S1210 and S1210.
Step 1211 is performed, and the process in step S114 ends. Then, as in the first data rebuilding process, the control unit 33 performs step S1 for all LBAs.
Steps S112 to S116 are executed until step S14 is executed.

According to the above-described second data restructuring process, the requests for data restructuring (the seventh read request and the sixth write request) are queued in the priority queue. Requests are not stuck in the request queue 36, thereby guaranteeing the completion time of the data reconstruction process. Furthermore, the second data rebuilding process is performed efficiently because the queuing and execution control of each request is performed for each disk device.

Next, the case where the host device 4 requests access to the LBA set as "rebuild required" in the new disk management table and the information recording device 1 has to execute reading. The operation will be described. In such a case, the host device 4 requests the reading of the data recorded in the LBA in which the status “rebuilding is required” is set, and the update data described with reference to FIG. 8 is read. Occurs when and.

The control unit 33 can confirm that the target LBA needs to undergo the data reconstructing process by referring to the table storage unit 311 at the time of the reading process. Recognize that the read data cannot be read. The control unit 33 accesses the table storage unit 311 and updates the status of the target LBA to “rebuilding”. And the control unit 33
Creates a read request for reading reproduction data necessary to reproduce the data of the target LBA, and queues it in a non-priority queue or a priority queue assigned to the disk device having the LBA. At this time,
If the control unit 33 is executing the read related to the first access request, the control unit 33 queues the read in the non-priority queue,
If the read related to the access request is being executed, it is queued in the priority queue.

Thereafter, the reproduction data is read out and stored in the buffer area in the buffer management section 310. The control unit 33 causes the parity calculation unit 39 to execute the parity calculation when all the data for reproduction are arranged in the buffer area, and reproduces the data of the target LBA. Control unit 33
Generates a seventh write request for writing the reproduced data to the target LBA while continuing the processing related to the read request from the host device 4 using the reproduced data, and In the non-priority queue for the disk device having The control unit 33
Accesses the table storage unit 311 at the time when the reproduced data is written, and updates the status of the target LBA to “normal state”.

Next, an operation in a case where the information recording apparatus 1 must execute writing to an LBA set as “rebuild required” in the new disk management table will be described. This operation is the same as the operation described with reference to FIG. 8, but when the control unit 33 creates a write request for each disk device, the table storage unit 311
And confirms that the status of the target LBA is “rebuilding required”, and then updates this status to “rebuilding”. Is completed, the target L
The difference is that the status of the BA is updated to the “normal state”.

As described above, when the host device 4 requests a read from the LBA set as “rebuild required” in the new disk management table, the information recording device 1 reads the data reproduced by the parity operation. , A process for writing to this LBA is executed. The fourth write request used for that is queued in the non-priority queue. Therefore, the reconstructed data, although non-prioritized,
The data is written to the disk array 2 in parallel with the processing according to the access request from the host device 4. As described above, the LBA set as “rebuild required” is subjected to the first or second data rebuild processing. However, since the number of LBAs set as “rebuild required” at the time of execution of each rebuild process is reduced by the parallel writing as described above, the time required for each rebuild process is reduced. You. Further, since the seventh write request is queued in the non-priority queue, such writing does not adversely affect the real-time processing to be executed in the information recording device 1.

When the host device 4 requests writing to an LBA set as “rebuild required” in the new disk management table, the status is updated to “normal state” at the same time as the completion of the writing. In addition, the information recording device 1 does not need to execute unnecessary reconstruction processing,
The processing time can be reduced.

Although the information recording apparatus 1 has the RAID3 and RAID4 architecture in the above embodiment, it may have the RAID5 architecture. Furthermore, the ratio of the number of disk devices that record parity blocks to the number of disk devices that record data blocks is 1: 1 and even if the disk has a RAID1 architecture that records the same block on both disks, This embodiment can be applied.

Although the case where there is one parity group has been described, the present invention can be similarly applied to a disk array 2 having a plurality of parity groups. In addition, the host device 4 uses the LUN to specify the priority. For example, when the first bit of the LBA is 1, the priority is indicated to a part of the LBA, such as setting the priority higher. Information may be added.

In the information recording apparatus 1, two types of priorities are set. However, three or more types of priorities may be set, and three or more types of queues corresponding to the priorities may be provided. . In this case, when the reconstructing process of the first data is performed, the request related to the reconstructing process created by the control unit 33 has a higher priority than the request related to the non-real-time process. By being queued as having a low priority, the reconstruction process can be performed without affecting the non-real-time process. Further, when the rebuilding process of the second data is performed, a request related to the rebuilding process created by the control unit 33 has a higher priority than a request created in connection with the real-time process. By being queued as a degree, the reconstruction process is not affected by the real-time process at all, and the completion time of the reconstruction process can be more strictly guaranteed.

When always requesting real-time processing or always requesting non-real-time processing, the host device 4 does not need to include information indicating the priority in the access request. No longer needed. In addition, the first and second reconstruction processes are described as two independent processes. However, the two processes are executed at the same time to guarantee the completion time of the reconstruction process. Thus, the reconstruction process is executed more efficiently.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an information recording device 1 according to an embodiment of the present invention.

2 is a diagram showing a SCSI message transmitted by the host device 4 shown in FIG. 1 as Identify and Simple Q
It is a figure showing the format of ueue Tag.

FIG. 3 is a diagram showing a format of a Read 10 and a Write 10 which are SCSI commands transmitted by the host device 4 shown in FIG.

FIG. 4 is a view showing a format of a processing request generated by a host interface unit 31 shown in FIG. 1;

FIG. 5 is a flowchart showing a processing procedure in the information recording device 1 when the host device 4 requests reading of non-real-time data or real-time data in the parity group “A”.

FIG. 6 is a diagram showing a format of a read request or a write request created by a control unit 33 shown in FIG.

FIG. 7 is a flowchart showing a detailed processing procedure of step S59 shown in FIG. 5;

FIG. 8 is a flowchart showing a processing procedure in the information recording device 1 when the host device 4 requests writing of non-real-time data or real-time data in the parity group “A”.

FIG. 9 shows first to fifth disk management tables 3111 to 31 stored in the table storage unit 311 shown in FIG.
FIG.

FIG. 10 shows each disk management table 3111 shown in FIG. 9;
It is a figure which shows the status set to ~ 3115.

FIG. 11 is a flowchart showing an overall procedure of a first data rebuilding process executed by the information recording apparatus 1;

FIG. 12 is a flowchart showing a detailed processing procedure of step S114 shown in FIG. 11;

FIG. 13 is a flowchart showing an overall procedure of a second data restructuring process executed by the information recording apparatus 1.

FIG. 14 is a flowchart showing a detailed processing procedure of step S114 shown in FIG.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Information recording device 2 ... Disk array 3 ... Array controller 33 ... Control part 341-345 ... 1st-5th non-priority queue 351-355 ... 1st-5th priority queue 36 ... Request queue 37 ... Request selection Unit 39: parity operation unit 4: host device

Claims (13)

[Claims] 1. A data input / output device for use in an information recording device including a redundant array comprising a recording medium for recording redundant data and an array controller for controlling the redundant array in accordance with an access request transmitted from a host device. The method, wherein the array controller creates a read request or a write request for the redundant array having a predetermined priority based on the input access request, wherein the array controller creates the created read request or write. Queuing a request in a queue therein, according to the predetermined priority, and the array controller selects, from the read request or the write request queued in the queue according to the predetermined priority, the redundant request. The array selects what to process, and the redundant array Processing the read request or write request that is selected by Lee controller, the data input-output method. 2. The array controller includes therein a queue corresponding to the priority, and the created read request or write request is queued in a queue corresponding to the predetermined priority. The data input / output method according to claim 1, wherein: 3. The array controller includes therein a queue corresponding to the predetermined priority for each recording medium, and the array controller determines a predetermined priority based on an input access request. Creating a read request or a write request for each recording medium, wherein the array controller queues the read request or the write request created for each recording medium in a queue of the corresponding recording medium in accordance with the predetermined priority. 2. The data input / output method according to claim 1, wherein the data input / output is performed. 4. The data input / output method according to claim 1, wherein said predetermined priority is set based on real-time processing of processing in said redundant array. 5. When the input / output interface with the host device conforms to SCSI, the predetermined priority is set to LUN or L of the access request.
2. The data input / output method according to claim 1, wherein the data is set in advance in a BA field. 6. An information recording apparatus which includes a redundant array formed of a recording medium for recording redundant data, and controls the redundant array in accordance with an access request transmitted from a host device. A selecting unit for selecting, from a read request or a write request queued in the queue, to be processed by the redundant array according to a predetermined priority, wherein the redundant array is selected by the selecting unit. An information recording device for processing a read request or a write request. 7. The request queue includes therein a queue corresponding to the priority, and a read request or a write request created by the control unit is queued in a queue corresponding to the predetermined priority. 7. The data input / output method according to claim 6 , wherein: 8. The request queue includes a queue corresponding to the predetermined priority for each of the recording media, and the control unit determines a predetermined priority based on an input access request. Creating a read request or a write request for each recording medium, wherein the read request or the write request created for each recording medium by the control unit is queued in a queue of the corresponding recording medium according to the predetermined priority. 7. The data input / output method according to claim 6, wherein the data input / output is performed. 9. An information recording apparatus comprising: a redundant array comprising a recording medium for recording redundant data; and an array controller for controlling the redundant array in accordance with an access request transmitted from a host device. A data reconstruction method for reproducing and reconstructing data recorded on a failed recording medium that is a failed recording medium, wherein the array controller has a predetermined priority. A read request or a write request for a necessary redundant array is created, and the array controller queues the created read request or write request in a queue therein, according to a predetermined priority. Among the queued queues according to the predetermined priority Selecting the created read request or write request, the redundant array processes the selected read request or write request, and the array controller causes the read request or write request to be processed by the redundant array. A data restructuring method for executing a data restructuring process based on the result obtained. 10. The array controller includes therein a queue corresponding to the predetermined priority for each of the recording media, and the array controller is required for data reconstructing processing and determines a predetermined priority. Creating a read request or a write request for each of the recording media, and queuing the read request or the write request created for each of the recording media in a queue of a corresponding recording medium according to the predetermined priority. 10. The data input / output method according to claim 9, wherein The 11. read and write requests created by the array controller, wherein the non-preferential processing priority as in the redundant array is subjected, according to claim 9 How to rebuild data. The 12. read and write requests created by the array controller, wherein the priority, as preferential treatment in the redundant array is subjected, according to claim 9 How to reconstruct the data. 13. A redundant array comprising a recording medium for recording redundant data, and an array controller for controlling the redundant array in accordance with an access request transmitted from a host device, and further comprising a failure in the redundant array. A data input / output method used in an information recording apparatus that reproduces data recorded on a failure recording medium that is a recording medium and reconstructs the spare recording medium, wherein the data to be reconstructed on the spare recording medium is provided. When an access request to the information recording device is transmitted from the host device to the information recording device, the array controller transmits, from the redundant array, reproduction data necessary for reproducing data recorded on the failure recording medium. Reading the array controller to reproduce the read from the redundant array A predetermined operation is performed using the data for reproduction to reproduce the data recorded on the failure recording medium, and the array controller has a predetermined priority for writing the reproduced data on the spare recording medium. The array controller queues the created write request in a queue having the same according to the predetermined priority, and the array controller according to the predetermined priority. While selecting the created write request from the queue as the one to be processed by the redundant array, the redundant array processes the write request selected by the array controller, and processes the reproduced data. Writing to the spare recording medium, wherein the write request is given a relatively low priority. That, data input and output method.