JP2000276305A - Disk array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform a recovering process without lowering the speed of a host I/O by allowing a load measuring means to measure the busy rates of respective disk drives and command the start and quitting of the recovering process according to the busy rates. SOLUTION: A host I/O processing means 120 receives a read/write request from a host 100, issues a corresponding indication to a disk drive 170, and returns its result to the host 100. A recovering process means 130 issues a read/ write indication to the disk drive 170 according to the report from the disk drive 170 and controls the load on the disk drive 170 according to the indication of the load control means 150 to perform the recovering process. In this case, a load control means 150 controls the selection of disk drives 170 used for read transfer from the host I/O processing means 120 and the implementation of the recovering process means 130 according to, for example, busy rates measured by a load measuring means 180.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array device, and more particularly, to a disk array device that stores the same data on a plurality of disks to recover data when a failure occurs.

[0002]

2. Description of the Related Art A disk array device adds redundant data to user data and stores the data in a distributed manner over a plurality of disk devices. Therefore, even if a failure occurs in one or a plurality of disk devices, the disk array device can store the user data. Can be restored. An example of such a conventional disk array device is disclosed in Japanese Patent Application Laid-Open No. 9-288546.

In the conventional recovery processing of the disk array device, if the recovery instruction from the host is a high-speed processing, the host I / O processing is performed for a short time, and if the recovery instruction is a low-speed processing, the host I / O processing is performed. Process for a long time. As a result, the recovery process and the host I / O process are controlled, and an efficient disk array device has been realized.

[0004]

However, in the above-mentioned conventional example, there is an inconvenience that the host I / O performance is remarkably deteriorated during restoration at high speed. Then, the operation of the entire device including the disk array device is reduced,
There may be a disadvantage that a required response time to the host device cannot be secured.

[0005]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a disk array device which can solve the disadvantages of the prior art and, in particular, can perform a recovery process while maintaining the host I / O performance. .

[0006]

SUMMARY OF THE INVENTION Therefore, the present invention comprises a plurality of disk devices, an array controller for controlling data transfer to the plurality of disk devices, and a host interface for connecting the array controller to a host device. ing. A recovery processing means for performing recovery processing of data to the disk device replaced by the occurrence of the failure by the array controller; and a host I / O for distributing a read / write command issued by the host device to the disk devices. Processing means, load measurement means for measuring the busy rate of each of the disk devices, and load control for instructing the recovery processing means to start and stop recovery processing based on the busy rate measured by the load measurement means. And means are provided. This aims to achieve the above-mentioned object.

In the present invention, the load measuring means measures the busy rate of each disk device, and the load control means instructs start and stop of the recovery processing based on the busy rate. The means stops the recovery process when the frequency of use by the recovery source or recovery destination host device is high, and resumes the recovery process when the access from the host device falls below a certain frequency.

In a preferred embodiment, the restoration processing means has a stage-based restoration processing control function of performing restoration processing in a plurality of stages in accordance with setting information of a predetermined restoration processing speed, and the load control means includes It is preferable to have a restoration processing speed update function for changing the setting of the restoration processing speed based on a busy rate to the host device.

In this example, the load control means changes the speed of the recovery processing according to the busy rate to the host device. That is, when I / O with the host device is frequently performed, a setting for reducing the recovery processing speed is performed.
Then, the recovery processing means performs the recovery processing at the recovery processing speed set by the load control means.

[0010]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the disk array device according to the present embodiment. As shown in FIG. 1, the disk array device according to the present embodiment includes a plurality of disk devices 170 and a plurality of disk devices 17.
An array controller that controls data transfer to the host device 0 and a host interface 115 that connects the array controller to a host device are provided. Then, the array controller performs recovery processing means 13 for recovering data to the disk device replaced by the occurrence of the failure.
0, a host I / O processing unit 120 for distributing a read / write command issued by the host device 100 to each of the disk devices # 1 to #n, and a busy rate for each of the disk devices # 1 to #n. And a load control unit 150 for instructing the recovery processing unit 130 to start and stop the recovery processing based on the busy rate measured by the load measurement unit 150.
And

The host I / O processing means 120
00, a corresponding instruction is issued to the disk device 170, and the result is sent to the host 10.
Return to 0. The recovery processing means 130
Read to disk device 170 based on report from
It issues a write instruction and controls the load on the disk device 170 according to the instruction from the load control means 150 to perform a recovery process.
The load measuring unit 180 measures the load of the disk device 170. This load measuring means has a function of, for example, measuring the connection time for each disk device 170 with respect to the host device 100 for each fixed period and calculating the host busy rate from the connection time for the fixed time. In this case, the load measuring unit 180 measures the elapsed time of accessing the disk device 170 for each fixed period, and measures a busy rate obtained by calculating a ratio of the fixed time to the access elapsed time. For this access, the host I / O processing means 12
0 and the recovery processing means 130. In the example shown in FIG. 1, the host busy rate for data transfer with the host device and the recovery processing busy rate for data transfer by the recovery processing are measured.

[0012] The load control means 150, based on, for example, the busy rate measured by the load measurement means, the host I / O
Disk device 1 used for read transfer from O processing means
70 and control the execution of the recovery processing means. For example, the load control unit 151 may have a function of controlling the start and stop of the recovery processing based on the busy rate of the magnetic disk device serving as the recovery source among the disk devices to the host device. And the disk device 17
0 performs data read / write operation and stores data.

Further, in the example shown in FIG.
30 includes a stage-based recovery process control function 131 that performs a recovery process in a plurality of stages according to the setting information of a predetermined recovery processing speed. The load control unit 150 includes a recovery processing speed update function 151 that changes the setting of the recovery processing speed based on the busy rate for the host device 100.
This is because, for example, when the busy rate of the recovery source or recovery destination disk to the host device exceeds a predetermined threshold value, a slow recovery process is selected, and on the other hand, the threshold value is not exceeded. In such a case, a high-speed recovery process is selected. Instead of such two-stage adjustment, a plurality of stages may be used. For example, the frequency at which recovery processing is executed is measured as a recovery processing busy rate, and when the recovery processing busy rate is exceeded, when processing to wait for recovery processing is performed, the threshold value of the recovery processing busy rate is set. Two types, one for high-speed recovery processing and one for low-speed recovery processing, are prepared, and switching between high-speed recovery and low-speed recovery is performed by selecting this threshold value.

In this embodiment, the load measuring means 18
0 and the load control means 151 operate asynchronously,
Load monitoring table 1 for storing the busy rate measured by load measuring means 180 in the local memory shown in FIG.
60 is provided. In the local memory, a host I / O busy rate threshold, a busy rate limit at the time of high-speed recovery,
The load control table 200 stores the value of the busy rate limit at the time of low-speed recovery and can change the value at any time. The array conversion table 140 is a table for converting a read / write request from the host device 100 into a physical address in the disk array device.

FIG. 2A shows an example of the load monitoring table. The load measuring unit 180 periodically calculates the busy rate of the load monitoring table. This processing will be described with reference to the flowchart of FIG. First, a new busy rate is recalculated from the disk usage time and the busy rate update time interval of the load monitoring table, and the busy rates of host I / O, recovery, and overall processing are updated (step A1 in FIG. 3). Subsequently, the disk usage time of the disk device 170 is initialized to 0 (step A2 in FIG. 3). This process is performed for all the disk devices, and after waiting for a fixed time (several seconds) (FIG. 3)
Steps A3 and A4) are repeated.

FIG. 4 is a flowchart showing an example of a specific busy rate calculation process. The load measuring means 180
Host I / O processing means 120 or recovery processing means 130
Saves the time when was connected to the disk device 170 (step B1 in FIG. 4). When the processing for the disk device (step B2 in FIG. 4) is completed, the disk device usage time is calculated from the time at that time and the time saved at the start of the processing (step B3 in FIG. 4). Further, the position of the table of the disk device 170 used in the transfer of the load monitoring table is obtained (step B in FIG. 4).
4). At this time, if the disk device 170 uses the host I / O processing means 120, the disk usage time of the host I / O is used. If the recovery processing means 130 is used, the disk usage time of the recovery processing means is used. The disk device usage time is added (steps B6 and B7 in FIG. 4). Then, the same value is added to the disk use time of the entire process (step B8 in FIG. 4).

Next, an example of processing when a read transfer request is issued from the host device will be described with reference to FIG. As shown in FIG. 5, when a read I / O command is instructed from the host 100, the host I / O processing unit 120 determines that the transfer is a read transfer (step C1 in FIG. 5). Then, referring to the array conversion table 140, the corresponding two disk device numbers and addresses are obtained (step C2 in FIG. 5). Next, the load control unit 150 recalculates the overall processing busy rate of the load monitoring table 160 from the table update time and the disk usage time, and obtains an average value with the current busy rate (step C3 in FIG. 5). Load control means 1
50 determines the disk device number with the low busy rate of the recalculated overall processing, and performs read transfer using the disk device 170 (step C4 in FIG. 5).

When a write I / O command is instructed from the host 100, it is determined that the transfer is a write transfer (step C1 in FIG. 5).
The disk device number and address corresponding to the ID1 array disk are obtained (step C5 in FIG. 5), and write transfer is performed to all the relevant disk devices 170 (step C6 in FIG. 5).

Next, the operation of the restoration process will be described with reference to FIG. When a certain disk device 170 is being recovered, the recovery processing unit 130 sets an address for recovery (step D1 in FIG. 6), and then the host I / O busy rate of the recovery source disk device 170 calculated by the load control unit 150. It is determined whether or not the restoration process is possible based on When the load control means 150 is instructed by the restoration processing means 130 to determine whether or not the processing can be executed, the load monitoring table (FIG. 2A)
The host I / O busy rate of the recovery source disk unit is checked from the host I / O ratio, and the value is used as the host I / O busy rate in the load control table (FIG. 2B).
Compare with the O busy rate threshold (Step D in FIG. 6)
3). If the host I / O busy rate is lower than the threshold value, the high-speed recovery busy rate limit of the load control table is set. If the host I / O busy rate is higher than the threshold value, the low-speed recovery busy of the load control table is disabled. Using the rate restriction, it is determined whether the recovery rate of the recovery source disk device 170 exceeds the restriction (steps D4 and D5 in FIG. 6).

If the busy rate of the recovery processing exceeds the limit, after waiting for a predetermined time (step D2 in FIG. 6), the load control means 150 checks again whether the recovery processing is possible or not. When the recovery processing busy rate does not exceed the limit, a read transfer is performed from the recovery source disk device 170 to the local memory 190 (step D6 in FIG. 6), and the data is written and transferred to the recovery destination disk device 170 (FIG. 6). Step D
7).

Subsequently, if it is not determined whether all the data has been restored, the address is updated (step D1 in FIG. 6), and the process is repeated. Otherwise, the restoration process is ended. The host I / O busy rate threshold, the fast recovery busy rate limit, and the low speed recovery busy rate limit in the load control table are set to appropriate values before the operation of the disk array device 110 is started. It is a value that can be changed.

As described above, according to the present embodiment, the host I / O and the busy rate of the recovery processing are periodically calculated, and the execution of the recovery processing is restricted by the busy rate. Performance degradation given to I / O can be set freely. Further, the busy rate of the disk device is calculated periodically, and the disk device having a lower busy rate is preferentially used for the read processing, so that the read performance can be improved.

[0023]

According to the present invention, the load measuring means measures the busy rate of each disk device, and the load control means measures the busy rate. The recovery processing means stops the recovery processing when the frequency of use by the recovery source or recovery destination host device is high, and the access from the host device falls below a certain frequency. Thus, it is possible to provide an unprecedented excellent disk array device in which the recovery process can be restarted when the error occurs and the recovery process can be performed without reducing the speed of the host I / O.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a first embodiment of the present invention.

2 is an explanatory diagram showing an example of the contents of a table shown in FIG. 1; FIG. 2 (A) is a diagram showing an example of a load monitoring table; FIG. 2 (B) is a diagram showing an example of a load control table; It is.

FIG. 3 is a flowchart showing an outline of a busy rate calculation process in the configuration shown in FIG. 1;

FIG. 4 is a flowchart illustrating an example of a detailed configuration of a busy rate calculation process illustrated in FIG. 3;

FIG. 5 is a flowchart illustrating a processing example when a read request is issued from a host device in the configuration illustrated in FIG. 1;

FIG. 6 is a flowchart illustrating an example of a recovery process in the configuration illustrated in FIG. 1;

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 host device 110 disk array device 115 host interface 120 host I / O processing means 130 recovery processing means 150 load control means 170 plural disk devices 180 load measurement means

Claims (6)

[Claims] 1. A disk array device comprising: a plurality of disk devices; an array controller for controlling data transfer to the plurality of disk devices; and a host interface for connecting the array controller to a host device. Recovery processing means for recovering data to the disk device replaced by the occurrence of the failure, host I / O processing means for distributing a read / write command issued by the host device to each of the disk devices, Load measuring means for measuring a busy rate of each disk device; and load control means for instructing the recovery processing means to start and stop recovery processing based on the busy rate measured by the load measuring means. A disk array device characterized by the above-mentioned. 2. The method according to claim 1, wherein the recovery processing unit has a recovery processing control function for performing a recovery processing in a plurality of stages in accordance with setting information of a predetermined recovery processing speed. 2. The disk array device according to claim 1, further comprising a recovery processing speed update function for changing the setting of the recovery processing speed based on the command. 3. The load control means has a function of controlling start and stop of the recovery process based on a busy rate of the magnetic disk device as a recovery source among the disk devices to the host device. Claim 1
The disk array device according to the above. 4. The apparatus according to claim 1, wherein the load measuring means has a function of measuring a connection time for the disk with respect to the host device for each fixed period and calculating a host busy rate from the connection time for the fixed time. The disk array device according to claim 1, wherein 5. The disk array device according to claim 4, wherein said load measuring means has a function of calculating an average of a busy rate calculated every said fixed period and a busy rate calculated last time. 6. The host I / O processing means has a function of selecting a disk having the lowest busy rate among a plurality of disk devices when a read command is received from the host device. The disk array device according to claim 1, wherein