JPH06250795A - Disk array system - Google Patents

Abstract

(57) [Summary] [Purpose] A disk array system that provides an operator with an index for determining how much priority should be given to copy processing when performing disk recovery processing, and performs optimal control based on that index. I will provide a. [Structure] The monitor unit 235 measures the input / output request arrival rate, the average response time, and the disk recovery time, and the monitor result 2
Record as 36. At the start or during the disk recovery, the index estimation unit 236 uses the current input / output request arrival rate and the monitor result 236 to determine the disk recovery time,
The average response time and average performance maintenance time are estimated and presented to the operator as a reference when designating the copy speed, or based on that, the optimum control is automatically performed to execute the copy processing. [Effect] Appropriate disk restoration processing can be performed from the viewpoint of both reliability and online processing performance.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array system, and more particularly to a disk array system capable of suitably performing disk recovery processing when a disk drive fails.

[0002]

2. Description of the Related Art A disk array system records data in a distributed manner on a plurality of disk drives, and when one or more disk drives fail, the data held by the failed disk drives is replaced by a normal disk drive. By performing a copy process that restores based on the data in the drive and writes it to the repaired / replaced disk drive or spare disk drive by several tracks, it is possible to perform the recovery process for the failure while continuing the online process. It is a system that Such a disk array system is, for example, "a disk array (Nikkei Computer 1992.3.
9; pp. 87-98) ".

[0003]

In the disk array system, if the priority of the copy process is increased, the time during which some of the disk drives are in a failed state is shortened. The probability of falling into a state where restoration is impossible is reduced. That is, the reliability is improved. On the other hand, the online processing becomes sparse by that amount, and the online processing performance decreases. That is, in a disk array system, online processing and copy processing are performed in parallel, but it is difficult to determine how much priority is given to copy processing compared to online processing. However, heretofore, no technology regarding the priority of the copy processing has been disclosed.

The present invention has been made in view of such circumstances, and an object thereof is to calculate an index for deciding how much priority should be given to copy processing when performing disk recovery processing. Then, it is to provide a disk array system that presents it to the operator and performs optimum control based on it, and performs appropriate disk recovery processing from the viewpoint of both reliability and online processing performance.

[0005]

According to a first aspect of the present invention, the present invention provides an input / output request arrival rate from a host computer, a disk recovery time until completion of copy processing, a copy speed, and an average during disk recovery processing. Monitoring means for measuring the I / O request arrival rate and recording them as a monitoring result, and a copy speed and disk recovery assuming that the I / O request arrival rate at the present time is maintained based on the monitoring result. The present invention is characterized by comprising index estimation means for obtaining a time relationship and outputting it as an estimation result, and service means for displaying the estimation result as an index for specifying a copy speed and for prompting an operator to specify the copy speed. A disk array system is provided.

In a second aspect, the present invention relates to an input / output request arrival rate from a host computer, a disk recovery time until completion of copy processing, a copy speed, and an average input / output request arrival rate during disk recovery processing. The disk recovery when it is assumed that the current I / O request arrival rate is maintained on the basis of the monitoring means for measuring and recording them as a monitor result and the copy speed designated by the operator and the monitor result. There is provided a disk array system comprising an index estimating means for obtaining time and outputting it as an estimation result and a service means for displaying the estimation result.

In a third aspect, the present invention provides an input / output request arrival rate from a host computer, a disk recovery time until completion of copy processing, a copy speed, and an average input / output request arrival rate during disk recovery processing. And a monitoring means for recording them as a monitor result, and a relationship between the copy speed and the disk recovery time when the current input / output request arrival rate is assumed to be maintained is obtained based on the monitor result. Further, based on the disk recovery time, the failure rate of one disk drive, and the repair replacement rate, the average time MTTDL until the data cannot be restored due to the failure of a plurality of disk devices is calculated and calculated. An index estimation unit that outputs the estimation result and a service unit that displays the estimation result as an index for designating a copy speed and causes an operator to designate the copy speed are provided. To provide a disk array system characterized by the.

In a fourth aspect, the present invention provides an input / output request arrival rate from the host computer, an average response time until completion of processing of the input / output request from the host computer during the disk restoration process, and a copy speed. It is assumed that the average I / O request arrival rate during the disk recovery process is measured and monitoring means for recording them and the I / O request arrival rate at the present time are maintained based on the monitoring result. The index estimation means for obtaining the relationship between the copy speed and the average response time and outputting it as an estimation result, and the service means for displaying the estimation result as the index for specifying the copy speed and allowing the operator to specify the copy speed. A disk array system characterized by the above.

In a fifth aspect, the present invention provides an input / output request arrival rate from a host computer, a response time until completion of processing of an input / output request from the host computer, and a disk recovery time until completion of copy processing. Measure the copy speed and the average I / O request arrival rate during disk recovery processing,
A monitoring means for recording them as a monitoring result, and a relationship between the copy speed and the disk recovery time when the present input / output request arrival rate is assumed to be maintained is obtained based on the monitoring result. Based on the target response time specified in the above and the monitoring result, the response time for each copy speed is estimated assuming that the I / O request arrival rate is maintained at the present time, and the response time is calculated from the target response time. If it is smaller, the performance is maintained, and if it is larger, the performance is degraded.
Based on the failure rate and repair / replacement rate of one disk drive and the status classification, the average performance that functions in the above performance maintaining state until the data cannot be restored due to the failure of multiple disk drives. An index estimating means for calculating the maintenance time and outputting it as an estimation result, and a service means for displaying the estimation result as an index for designating a copy speed and allowing an operator to designate the copy speed are provided. To provide a disk array system for performing.

According to a sixth aspect of the present invention, in the disk array system having the above structure, the monitor means checks the copy processing execution status and records it in the monitor result, and the index estimation means, during the copy processing execution, The remaining time required for disk recovery is estimated based on the monitor result and recorded in the estimated result, and the service means displays the estimated result as a copy speed change instruction index during the execution of the copy process to the operator. There is provided a disk array system characterized by comprising a copy processing means for instructing a change of a copy speed, and further, when a copy processing operator gives a copy speed change instruction, changing the copy speed to perform a copy process. .

According to a seventh aspect of the present invention, in the disk array system having the above construction, the monitor means checks the copy processing execution status and records it in the monitor result, and the service means receives the target response time from the operator. When the average response time is smaller than the target response time based on the monitoring result and the target response time during the execution of the copy process, the copy speed is changed to a larger value or the frequency of the copy execution is increased. If the average response time is longer than the target response time, the copy speed is changed to a lower speed or the copy execution frequency is changed to a lower value, and the copy condition is changed by the copy adjusting means. There is provided a disk array system characterized by comprising a copy processing means for changing a copy condition and performing a copy processing when .

According to an eighth aspect of the present invention, in the disk array system according to the seventh aspect, when a spare disk drive is provided and a target response time and a copy speed at the start of copying are given in advance, A disk array system characterized in that each means starts an operation for a copy process without waiting for an operator's instruction.

According to a ninth aspect, the present invention provides an input / output request arrival rate from a host computer, a response time until completion of processing of an input / output request from the host computer, and a disk recovery time until completion of copy processing. Measure the copy speed and the average I / O request arrival rate during disk recovery processing,
A monitoring means for recording them as a monitoring result, and a relationship between the copy speed and the disk recovery time when the present input / output request arrival rate is assumed to be maintained is obtained based on the monitoring result. Based on the target response time specified in the above and the monitoring result, the response time for each copy speed is estimated assuming that the current I / O request arrival rate is maintained, and the average response time is the target response time. If it is smaller, the performance is maintained, and if it is larger, the performance is deteriorated, and a plurality of units are classified based on the disk recovery time, the failure rate of one disk drive, the replacement rate, and the status classification. Calculate the average performance maintenance time that is functioning in the above performance maintenance state until the data cannot be restored due to the failure of the disk drive, and use it as the estimation result. Index estimating means for inputting, a copy adjusting means for referring to the estimation result and setting a copy speed that maximizes the average performance maintaining time, and a copy processing for starting the copy processing at the copy speed set by the copy adjusting means The present invention provides a disk array system comprising:

According to a tenth aspect of the present invention, in the disk array system according to the ninth aspect, the monitor means measures the response time and also the number of tracks for which copying is completed and the elapsed time from the start of copying. Then, the index estimation means re-estimates the average performance maintenance time for each copy speed based on the target response time designated by the operator and the monitoring result while the copying process is being executed. This is set as an estimation result, the copy adjusting means refers to the estimation result at an appropriate cycle, and changes it to a copy speed that maximizes the average performance maintenance time. There is provided a disk array system characterized in that when there is a change, the copy speed is changed to perform copy processing.

According to an eleventh aspect of the present invention, in the disk array system according to the ninth or tenth aspect, the present invention provides an operator when a spare disk drive is provided and a target response speed is given in advance. There is provided a disk array system characterized in that each means starts an operation for copy processing without waiting for the instruction.

[0016]

In the disk array system of the present invention according to the first aspect described above, as an index for determining how much priority should be given to the copy processing, the copy is performed assuming that the I / O request arrival rate at the present time is maintained. Present the relationship between speed and disk recovery time to the operator. Therefore, the operator can specify a copy speed that provides an appropriate disk recovery time from the viewpoint of both reliability and online processing performance.

In the disk array system according to the second aspect of the present invention, the disk recovery time is assumed when the copy processing is performed at the copy speed designated by the operator and the current I / O request arrival rate is maintained. , Present to the operator. Therefore, the operator can confirm whether or not the copy speed that gives an appropriate disk recovery time is specified from the viewpoint of both reliability and online processing performance.

In the disk array system of the present invention according to the third aspect, the operator is given the average time MTTDL until data cannot be restored as an index for deciding how much priority should be given to copy processing. Present. Therefore, the operator can specify a copy speed that provides an appropriate disk recovery time from the viewpoint of both reliability and online processing performance.

In the disk array system of the present invention according to the above fourth aspect, the copy is performed assuming that the current I / O request arrival rate is maintained as an index for deciding to what extent the copy processing should be prioritized. Present the relationship between speed and average response time to the operator. Therefore, the operator can specify a copy speed that provides an appropriate average response time in terms of both reliability and online processing performance.

In the disk array system of the present invention according to the fifth aspect, when the response time for each copy speed is shorter than the target response time, it is called a performance maintenance state, and the performance is maintained until data cannot be restored. When the total time that functions in the maintenance state is called the average performance maintenance time, the average performance maintenance time is presented to the operator as an index for determining how much priority should be given to copy processing. Therefore, the operator can specify a copy speed that provides an appropriate average performance maintenance time from the viewpoint of both reliability and online processing performance.

In the disk array system of the present invention according to the sixth aspect, while the copy processing is being executed, the remaining time required for the disk recovery is estimated and presented to the operator to instruct the copy speed to be changed to perform the copy processing. The copy process is continued at the copy speed changed by the operator. Therefore, it becomes possible to deal with the case where the input / output request arrival rate fluctuates during execution of the copy process.

In the disk array system according to the seventh aspect of the present invention, the copy speed or the copy execution frequency is automatically adjusted so that the average response time matches the target response time during the copy processing. Therefore, it becomes possible to deal with the case where the input / output request arrival rate fluctuates during execution of the copy process.

In the disk array system of the present invention according to the eighth aspect, when the disk array system of the present invention has a spare disk drive, without waiting for the operator's instruction, at a predetermined copy speed at the start of copying, Each means starts the operation for the copy process. Therefore, the operator's instruction is unnecessary, and the maintenance can be automated.

In the disk array system of the present invention according to the ninth aspect, the copy speed is automatically set and the copy process is started so that the average performance maintenance time is maximized.
Therefore, the operator's instruction becomes unnecessary. In the disk array system of the present invention according to the tenth aspect, the copy speed is automatically adjusted so that the average performance maintenance time is maximized during the copy processing. Therefore, it becomes possible to deal with the case where the input / output request arrival rate fluctuates during execution of the copy process. In the disk array system of the present invention according to the eleventh aspect described above, when the spare disk drive is provided, each means is set at a predetermined copy speed at the start of copying without waiting for the operator's instruction. Start the operation for the copy process. Therefore, the operator's instruction is unnecessary, and the maintenance can be automated.

[0025]

Embodiments of the present invention will now be described in detail with reference to the drawings. The present invention is not limited to this. FIG. 1 is a block diagram of a hardware configuration common to disk array systems according to first to ninth embodiments described later. This disk array system 1
00 is a host computer 1, a disk controller 2,
It is composed of disk drives 31 to 34, a service processor 4 and a console 5. In this embodiment,
It is assumed that the data is distributed and written to three of the four disk drives 31 to 34, and the error correction code is written to the remaining one. However, in the present invention, the number of disk drives is not limited, and m disk drives may be prepared and n of them may be error correction drives (m and n are arbitrary natural numbers satisfying m> n ≧ 1). ). The disk controller 2 is composed of a channel 21, a buffer memory 22, control processors 25 and 26, local memories 27 and 28, and a common memory 29. The first to ninth embodiments have different software configurations, that is, functions. Next, the first to ninth embodiments will be described in order.

-First Embodiment- FIG. 2 is a block diagram of a disk array system 100A according to the first embodiment of the present invention. Disk controller 2
Has a channel 21 and a buffer memory 22. A data transfer path P is provided between the host computer 1 and the channel 21, between the channel 21 and the buffer memory 22, and between the buffer memory 22 and the disk drives 31 to 34.
There are 1, P2 and P3.

The respective units 231 to 237 surrounded by squares in the disk control device 2 represent the functions realized by the microprograms of the control processors 25 and 26. One control processor may be assigned to each function, or only one control processor may have all the functions. Further, each piece of information 241 to 249 enclosed by an ellipse in the disk control device 2 is information which is exchanged between the above respective functions. Among these pieces of information, the information to be passed between the functions assigned to the different control processors 25 and 26 is the respective control processors 25 and 26.
It is placed in a common memory 29 that can be referred to and updated by. Further, information to be passed between the functions assigned to the same control processor is placed in the local memory 27 or 28 of the control processor.

The input / output request accepting unit 231 allows the host computer 1 to input / output requests (data read command / data read / write command) to the disk drives 31 to 34 via the channel 21.
When a data write command) is issued, it is detected and registered in the input / output processing waiting queue 241.

When the input / output processing execution right 242 is given from the dispatch control section 234, the input / output processing section 232 processes the input / output request registered at the head of the input / output processing waiting queue 241. When the input / output request is a data read command, if the three disk drives storing the data are normal, the three disk drives are accessed to read the data. Then, if one unit has failed, the data and error correction code are read from the remaining three units, the data stored in the failed disk drive is restored, and then, via the buffer memory 22, Transfer to the host computer 1. When the input / output request is a data write command, the data is divided into three, an error correction code is created, and each is written to a normal disk drive via the buffer memory 22.

When one of the four disk drives fails, the copy processing section 233 repairs / replaces it and then executes the copy processing. FIG. 3 shows a flow chart of the copy process. First, the process loops until the copy start instruction 244 is received from the service processor communication unit 237 (step 1000). Upon receiving the copy start instruction 244, the copy control information 245 including the disk drive number of the recovery target and the number of tracks handled in one copy process (hereinafter, referred to as copy speed) is read (step 1010).
Next, the track number for the copy process is initialized (step 1020), and the process is looped until the dispatch control unit 234 gives the copy process execution right 243 (step 1030).

When the copy processing execution right 243 is given,
The top track number to be copied is read out (step 1040). Next, the data is read from three normal disk drives, and the buffer memory 22
(Step 1050). The tracks from which data is read are the number of tracks indicated by the copy speed starting from the track number.

Next, the data of the disk drive to be restored is restored from the data read to the buffer memory 22 and stored in the buffer memory 22 (step 106).
0). Then, the restored data is transferred from the buffer memory 22 to the disk drive to be restored (step 1
070).

Next, the head track number of the next copy processing target is updated (step 1080). Finally, it is judged whether or not the disk recovery processing is completed (step 10
90) If not completed, return to step 1030. When it is completed, copy processing completion is set in the copy execution status 246 (step 1100), and the process returns to step 1000. The input / output processing unit 232 resets the input / output processing execution right 242 after detecting the input / output processing execution right 242. In addition, the copy processing unit 233
After detecting the copy process execution right 243, the copy process execution right 243 is reset.

The dispatch control unit 234 gives the input / output processing execution right 242 to the input / output processing unit 232 by an appropriate schedule process. In addition, the copy processing execution right is given to the copy processing unit 233. FIG. 4 shows a flow chart of the schedule processing. In this schedule process, normally, the input / output process execution right 242 is set, and the copy process execution right is set at regular intervals. First, until the copy start instruction 244 is received from the service processor communication unit 237, the input / output processing execution right 242 is set (steps 2000 and 2010).

When the copy start instruction 244 is received from the service processor communication unit 237, the timer is reset to 0 (step 2020). Then, the timer is checked, and if the predetermined set time (for example, 50 ms) has not elapsed, the input / output processing execution right 242 is set (steps 2030 and 2040). When the predetermined set time has elapsed, the copy processing execution right 243 is set (step 20).
50).

If the copy execution status 246 does not indicate that the copy process is completed, the above step 2020 is executed.
Return to step 2000, and if set, return to step 2000 (step 2060). Note that each processing execution right 242, 2
43 setting processing (steps 2010, 2040, 205
In 0), the process loops on the spot until the previously set process execution right is reset.

The monitor section 235 constantly checks the input / output requests from the host computer 1 entering the channel 21 and measures the input / output request arrival rate. Then, the input / output request arrival rate is recorded in the monitor result 247. The monitor result 247 shows the disk recovery time from the setting of the copy start instruction 244 to the completion of the copy processing in the copy execution status 246, the copy speed, and the average arrival rate of I / O requests during the disk recovery processing. Record. The disk recovery time, copy speed, and average I / O request arrival rate are monitored each time disk recovery processing is performed.
Statistical processing is performed to obtain the relationship shown in FIG.

When the index estimation unit 236 receives the estimation instruction 248 from the service processor communication unit 237, the index estimation unit 236 reads the input / output request arrival rate at that time from the monitor result 247, and it is assumed that the input / output request arrival rate is maintained. The relationship between the copy speed and the disk recovery time at that time is obtained from the relationship shown in FIG. 5, and the result is set as the estimation result 249. The relationship between the copy speed and the disk recovery time at a certain I / O request arrival rate is as shown in FIG.

The service processor communication section 237 has the following three functions. First, when the service processor 4 instructs the disk recovery process to start, the disk recovery time estimation instruction 248 is set, and when the estimation result 249 is set, it is notified to the service processor 4.
Secondly, when the copy control information (recovery target disk drive, copy speed) is specified by the service processor 4, the copy start instruction 244 and the copy control information 245 are set. Thirdly, when the copy processing completion is set in the copy execution status 246, the disk recovery processing completion is notified to the service processor 4.

The service processor 4 issues an instruction input by the operator via the console device 5 to the disk control device 2.
And the information sent from the disk control device 2 is displayed on the console device 5.

FIG. 7 is a sequence diagram of a disk restoration process using the above-mentioned functions. Service processor 4
When receiving a disk recovery instruction from the operator (step 3000), notifies the disk controller 2 of this (step 3010).

On the other hand, in the disk control device 2,
The index estimation unit 236 estimates the disk recovery time for each copy speed (step 3020), and then reports it to the service processor 4 (step 3030). The service processor 4 displays the estimation result of the disk recovery time for each copy speed on the console device 5 (step 3040). Then, when the operator designates a desired copy speed, the operator is notified of it (step 3050).

In the disk control device 2, the copy processing unit 233 performs the copy processing according to the notified copy speed (step 3060). After the copy processing is completed, the disk recovery processing completion is reported to the service processor 4 (step 3070). The service processor 4 displays a copy processing completion message on the console device 5 to notify the operator (step 3080).

If the operator designates the copy speed at the same time as instructing the disk restoration process, the index estimating unit 2
36 sets the estimated value of the disk recovery time for the designated copy speed in the estimation result 249.

According to the first embodiment described above, when the operator instructs the disk recovery process, the estimated value of the disk recovery time for each copy speed is displayed.
The operator can specify the copy speed in consideration of the probability that the second disk drive will fail, which has the effect of facilitating a disk recovery work plan. The functions of the index estimation unit 236 and the monitor unit 235 are transferred to the service processor 4 and the host computer 1, and the monitor unit 23 is used.
A dedicated measuring instrument having the function of 5 may be provided.

-Second Embodiment- In the first embodiment, the disk recovery time is estimated and displayed.
In the second embodiment, the average time until data restoration becomes impossible due to the failure of a plurality of (two in this embodiment) disk drives is estimated and displayed. This is realized by changing the function of the index estimation unit 236 of FIG. 2 as follows. The index estimation unit 236 analyzes the state transition diagram of the disk array system. The state transition diagram in the case of the present embodiment is as shown in FIG. That is, it has the following four states and six transitions. Status S0: Four disk drives are operating normally Status S1: One disk drive is out of order and the remaining three are operating normally Status S2: One disk drive is being recovered, The remaining 3 units are operating normally. State S3: Data restoration is impossible (system down) due to a failure of 2 disk drives. Transition (S0, S1): 1 of 4 disk drives.
Platform failure Transition (S1, S2): Repair / replace failed disk drive Transition (S2, S0): Disk drive recovery processing complete Transition (S2, S1): Recovery target disk drive failed Transition (S1, S3) : One of the three normal disk drives fails. Transition (S2, S3): One of the three normal disk drives fails. In Fig. 8, the label on each transition represents the transition rate. . λ represents the failure rate of one disk drive. μ represents the repair and replacement rate of one disk drive. T represents the disk recovery time. The values of the failure rate λ and the repair and replacement rate μ are given to the system in advance. Assuming that they follow an exponential distribution, the following differential equation holds based on the theory of Markov processes. Note that Pi (t) is in state S at time t.
Probability of existing in i.

[0047]

[Equation 1]

When the index estimation unit 236 receives the estimation instruction 248 from the service processor communication unit 237, the input / output request arrival rate at that time is read from the monitor result 247, and it is assumed that the input / output request arrival rate is maintained. The disk recovery time for each copy speed at that time is obtained from the relationship shown in FIG. Next, the values of λ, μ and T are substituted into the above (Equation 1) to solve the differential equation. Then, the average time until data restoration becomes impossible MTTDL (Mean Time To
Data Lost) is calculated by the following equation, and the relationship between each obtained copy speed and the average time MTTDL is set in the estimation result 249.

[0049]

[Equation 2]

The relationship between each copy speed and the average time MTTDL is displayed on the console device 5 as in the first embodiment.

According to the second embodiment described above, when the operator gives an instruction for the disk restoration process, the estimated value of the average time MTTDL value until the data restoration becomes impossible for each copy speed is displayed. With that in mind, the operator can specify the copy speed, which has the effect of facilitating a disk recovery work plan.

-Third Embodiment- In the first embodiment, the estimated value of the disk recovery time is displayed, and in the second embodiment, the estimated value of the average time MTTDL value until the data cannot be restored is displayed. In the third embodiment,
Estimate and display the response time from receiving the request to finishing the process. The configuration of the third embodiment is the same as that of the first and second embodiments except for the functions of the monitor unit 235 and the index estimation unit 236.

The monitor section 235 constantly monitors the signal passing through the channel 21, and checks the input / output request from the host computer 1 and the processing completion report sent to the host computer 1 by the input / output processing section 232. Then, the input / output request arrival rate and the response time from the reception of the request to the end of the processing are measured. The measured input / output request arrival rate and response time are recorded in the monitor result 247. Furthermore, the monitor unit 235 records the average I / O request arrival rate and the average response time in the monitor result 247 together with the copy speed at each time of one disk restoration process. The copy rate, the average input / output request arrival rate, and the average response time are statistically processed to obtain the relationship shown in FIG.

When the index estimating unit 236 receives the average response time estimation instruction 248 from the service processor communication unit 237, it reads the input / output request arrival rate at that time from the monitor result 247, and the input / output request arrival rate is The relationship between the copy speed and the average response time when it is assumed to be maintained is obtained. Then, the result is set as the estimation result 249. The relationship between the copy speed and the average response time at a certain I / O request arrival rate is as shown in FIG.

According to the third embodiment described above, when the operator instructs the disk recovery processing, the estimated value of the response time for each copy speed is displayed, so the online processing performance during the disk recovery processing is taken into consideration. Then, the operator can specify the copy speed, which has the effect of facilitating a disk recovery work plan.

Fourth Embodiment As can be seen by comparing FIGS. 6 and 10, the disk recovery time (reliability) and average response time (online processing performance)
Has a trade-off relationship with the copy speed. In the first to third embodiments, the operator specifies the copy speed by abstractly considering these balances, but in the fourth embodiment, the balance between reliability and online processing performance is quantitatively determined. An index called the average performance maintenance time is estimated and displayed.

In a degradable / restorable system such as a disk array system, among operating states, a state in which the performance exceeds the target performance (performance maintenance state) and a state in which it falls below the target performance (performance degradation state) ) There is. Therefore, the total time during which the system has been functioning in the performance maintaining state is considered as one index until the performance maintaining state and the performance deterioration state are alternately repeated until the final return to the performance maintaining state becomes impossible. This index is used as the average performance maintenance time THP (Total
High Performance Time). The average performance maintenance time THP is calculated by the following equation. SH represents a set of performance maintaining states.

[0058]

[Equation 3]

The monitor unit 235 monitors the input / output request arrival rate, the response time and the disk recovery time in each of the system states S0, S1, S2, S3, and records them in the monitor result 247.

The index estimation unit 236 receives the target response time at the same time as receiving the instruction 248 for estimating the average performance maintenance time THP from the service processor communication unit 237. This target response time is also input when the operator instructs the service processor 4 to restore the disk. Next, in the index estimation unit 236, the monitor result 24
Based on No. 7, each system state S0, S1, S assuming that the I / O request arrival rate at that time is maintained
2. Estimate the response time for each copy speed in S3. Then, by comparing the estimated response time for each copy speed with the target response time, each system state S0, S1, S2, S3 is classified into either a performance maintenance state or a performance degradation state. Next, the index estimation unit 236 obtains the disk recovery time for the I / O request arrival rate at that time based on the monitor result 247. Next, the values of λ, μ and T are substituted into the above (Equation 1) to solve the differential equation. Further, using the result and the classification result, the average performance maintenance time THP for each copy speed is calculated from the above (Equation 3).

Finally, the index estimation unit 236 sets the relation between the copy speed and the average performance maintenance time THP in the estimation result 249.

According to the fourth embodiment described above, when the operator instructs the disk recovery processing, the estimated value of the average performance maintenance time THP for each copy speed is displayed. The operator can specify the copy speed by quantitatively grasping the balance of the above, which has the effect of facilitating a disk recovery work plan.

-Fifth Embodiment- A fifth embodiment is an embodiment in which the operator can change the copy speed during the disk restoration process.

In addition to setting the copy processing completion in the copy execution status 246 when the copy processing is completed, the copy processing unit 233 copies the number of tracks that have been copied each time one copy processing is completed. The status 246 is set. The monitor unit 235 constantly checks the copy execution status 246, and sets the number of tracks for which copying has ended in the monitoring result 247 together with the elapsed time from the start of copying.

In the index estimation unit 236, the monitor result 247 is displayed.
From the above, the execution status from the start of copying is judged, and the remaining time until the copying process is completed is estimated. Then, the estimated remaining time is set in the estimation result 249. This processing may be always carried out spontaneously, or may be carried out when it is detected that the request from the operator is made by referring to the estimation instruction 248.

The service processor communication unit 237 notifies the service processor 4 of the estimation result 249 and sets it in the copy control information 245 when the operator gives a new instruction of the copy speed. Copy processing unit 23
In step 3, each time the copy process execution right 243 is obtained and the copy process is executed, the copy speed is read from the copy control information 245, and the copy process is executed at the instructed copy speed.

According to the fifth embodiment described above, during the disk recovery process, the remaining time until the copy process is completed is displayed and the copy speed can be changed, so the I / O request arrival rate during the disk recovery process. Even if the value changes, the operator can flexibly deal with the situation.

-Sixth Embodiment- A sixth embodiment is an embodiment in which the response time is monitored and the copy speed is automatically adjusted. FIG. 11 is a configuration diagram showing a disk array system 100B of the sixth embodiment. The monitor unit 235 constantly measures the response time and sets it as the monitor result 247. The service processor communication unit 237 also receives the target response time when receiving the disk recovery instruction from the service processor 4. Then, the target response time is set to the target value 250 and the copy start instruction 244 is set.

In the copy adjusting unit 238, the monitor result 24
Always read the response time from 7 and set it as the target value 250
Compare with. When the target value is larger, it is possible to give priority to the copy process, so the copy control information 24
Set a higher copy speed to 5. On the other hand, when the target value is smaller, it is necessary to suppress the copy process, so a smaller copy speed is set in the copy control information 245.

In the sixth embodiment described above, even if the I / O request arrival rate changes during the disk recovery processing, the copy processing can be automatically adjusted and the target performance can be maintained without operator intervention. . In addition to changing the copy speed, the priority of the copy process can be changed by changing the interval at which the dispatch control unit 234 sets the copy process execution right 243. Therefore, the same effect as described above can be obtained. Therefore, the copy adjusting unit 238 may automatically change the time interval for setting the copy processing execution right 243.

-Seventh Embodiment- The seventh embodiment is an embodiment for automatically setting / automatically adjusting the copy speed by utilizing the average performance maintaining time THP. Figure 1
2 is a block diagram showing a disk array system 100C of the seventh embodiment. First, similarly to the fourth embodiment, the average performance maintenance time THP for each copy speed is calculated, the copy speed that maximizes the average performance maintenance time THP is automatically set, and the copy process is started.

The monitor section 235 constantly measures the response time, and also measures the number of tracks for which copying has ended and the elapsed time from the start of copying, and sets this as the monitoring result 247. The index estimating unit 236 compares the response time read from the monitor result 247 with the target response time (target value 250) to classify the current state during execution of the disk restoration process into the performance maintenance state or the performance deterioration state. . Also, based on the monitor result 247, the disk recovery time T
To recalculate. Next, using the recalculated disk restoration time T, the average performance maintenance time THP is recalculated. Then, the average performance maintenance time THP is set in the estimation result 249.

The copy adjusting unit 238 refers to the estimation result 249 at an appropriate cycle, and compares it with the average performance maintenance time T from the previous time.
When the numerical value of HP deteriorates, the average performance maintenance time THP when the copy speed is changed is re-estimated.
The index estimation unit 236 is instructed. The index estimation unit 236 newly sets the average performance maintenance time TH for each copy speed.
P is re-estimated and set to the estimation result 249. The copy adjustment unit 238 obtains the copy speed at which the average performance maintenance time THP is maximum based on the estimation result 249, and sets it in the copy control information 245.

In the above seventh embodiment, even if the I / O request arrival rate changes during the disk recovery process, copying is performed so that the balance between reliability and online processing performance is optimized without operator intervention. The speed can be adjusted automatically.

The seventh embodiment may be changed to any of the following depending on the capability of the system. The copy adjustment unit 238 uses the estimation result 249 at an appropriate cycle.
When the value of the average performance maintenance time THP is worse than the average performance maintenance time THP calculated at the start of the copy process, the average performance maintenance time THP when the copy speed is changed is re-estimated. , And issues an instruction to the index estimation unit 236. The index estimation unit 236 compares the response time read from the monitor result 247 with the target response time (target value 250) to classify the current state during the disk recovery process into the performance maintenance state or the performance degradation state. , And set it as the estimation result 249. The copy adjustment unit 238 refers to the estimation result 249 at an appropriate cycle, and when the performance is degraded, the average performance maintenance time TH when the copy speed is changed
The index estimation unit 236 is instructed to re-estimate P. The index estimation unit 236 uses the monitor result 247 at an appropriate cycle.
, The disk recovery time T is recalculated. Next, using the recalculated disk recovery time T, the average performance maintenance time THP for each copy speed is recalculated. Then, the average performance maintenance time THP is set in the estimation result 249.

Eighth Embodiment A service processor may be connected to the host computer 1 so that a disk recovery start instruction from an operator reaches the disk controller 2 from the host computer 1 via the service processor. For example, the operator can instruct the start of disk recovery from the host computer 1. Further, even if the host computer 1 is provided with the function of the service processor, the operator can instruct the host computer 1 to start the disk recovery.

-Ninth Embodiment-In the first to eighth embodiments, the operator instructs the start of disk recovery after repairing or replacing the failed disk drive, but in the ninth embodiment Automatically starts disk recovery processing without waiting for disk drive repair or replacement. FIG. 13 is a configuration diagram showing a disk array system 100D of the ninth embodiment. The disk array system 100D has five disk drives 31 to 35. Of these, the disk drive 35 is a spare disk drive and is not accessed when the system is operating normally.

Further, the disk controller 2 is provided with a disk failure detecting section 239. The disk failure detection unit 239 has a function of detecting a disk failure during execution of input / output processing. When a disk failure is detected, the copy start instruction 244 is set and the failed disk drive number is written in the failure information 251.

When the copy start instruction 244 is set, the copy adjusting section 238 refers to the target response time and the monitor result 247 registered in advance in the target value 250,
Similarly to the sixth embodiment, the copy control information 245 is set. In the copy processing unit 233, the disk drives 31 to 3 are based on the failure information 251 and the copy control information 245.
Read data from 3 normal disks out of 4,
The restored data is written to the spare drive 35.

In the ninth embodiment described above, even if one disk drive fails, the disk recovery processing is automatically started without waiting for the repair / replacement of the disk drive, and the preset target performance is achieved. The disk recovery process can be executed while protecting it. In addition, instead of the response time,
The disk restoration process may be executed using the average performance maintenance time THP as in the embodiment.

[0081]

According to the disk array system of the present invention, an index for deciding how much priority should be given to the copy process when performing the disk recovery process is calculated and presented to the operator. Based on this, optimum control can be performed. Therefore, it becomes possible to perform appropriate disk restoration processing from the viewpoint of both reliability and online processing performance.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a disk array system according to an embodiment of the present invention.

FIG. 2 is a configuration diagram of a disk array system according to the first embodiment of this invention.

FIG. 3 is a flowchart of copy processing.

FIG. 4 is a flowchart of a schedule process.

FIG. 5 is a relationship diagram of an average input / output request arrival rate, a copy speed, and a disk recovery time.

FIG. 6 is a relationship diagram between copy speed and disk recovery time.

FIG. 7 is a flowchart of a sequence of disk restoration processing.

8 is a state transition diagram of the disk array system of FIG.

FIG. 9 is a relationship diagram of an average input / output request arrival rate, a copy speed, and an average response time.

FIG. 10 is a relationship diagram between copy speed and average response time.

FIG. 11 is a configuration diagram of a disk array system of a sixth embodiment of the present invention.

FIG. 12 is a configuration diagram of a disk array system of a seventh embodiment of the present invention.

FIG. 13 is a configuration diagram of a disk array system of a ninth embodiment of the present invention.

[Explanation of symbols]

1 ... Host computer, 2 ... Disk control device, 31
~ 35 ... Disk drive, 4 ... Service processor,
5 ... Console device, 21 ... Channel, 22 ... Buffer memory, 231 ... I / O request accepting unit, 232 ... I / O processing unit, 233 ... Copy processing unit, 234 ... Dispatch control unit, 235 ... Monitor unit, 236 ... Index estimation Part, 237
... service processor communication unit, 238 ... copy adjustment unit,
239 ... Disk failure detection unit, 241 ... Input / output processing waiting queue, 242 ... Input / output processing execution right, 243 ... Copy processing execution right, 244 ... Copy start instruction, 245 ... Copy control information, 246 ... Copy execution status, 247 ... Monitor result, 248 ... Estimation instruction, 249 ... Estimation result, 250 ... Target value, 251 ... Failure information.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Akira Kurano 2880 Kozu, Odawara City, Kanagawa Stock Company, Hitachi Ltd. Storage Systems Division

Claims (11)

[Claims] 1. A host computer, a disk controller, and a plurality of disk drives, which record data in a distributed manner in the plurality of disk drives, and when one or more disk drives fail, hold the failed disk. By restoring the data that was restored based on the data in the normal disk drive and writing it to the repaired / replaced disk drive or spare disk drive at the copy speed specified by the operator. In a disk array system that performs recovery processing against the above-mentioned failure while continuing online processing, the arrival rate of I / O requests from the host computer, the disk recovery time until the completion of the copy processing, the copy speed, and the disk recovery processing Measure the average I / O request arrival rate,
Monitoring means for recording them as a monitoring result, and based on the monitoring result, the relationship between the copy speed and the disk recovery time when the current I / O request arrival rate is assumed to be maintained is determined, and the estimation result is calculated. A disk array system, comprising: an index estimating unit that outputs as an index, and a service unit that displays the estimation result as an index for specifying a copy speed and allows an operator to specify the copy speed. 2. A host computer, a disk controller, and a plurality of disk drives, which record data in a distributed manner in the plurality of disk drives, and when one or more disk drives fail, hold the failed disk. By restoring the data that was restored based on the data in the normal disk drive and writing it to the repaired / replaced disk drive or spare disk drive at the copy speed specified by the operator. In a disk array system that performs recovery processing for the above-mentioned failure while continuing online processing, the arrival rate of I / O requests from the host computer, the disk recovery time until the completion of copy processing, the copy speed, and the disk recovery process Measure the average I / O request arrival rate,
Based on the monitoring means for recording them as a monitor result, the copy speed designated by the operator, and the monitor result, the disk recovery time is calculated assuming that the current I / O request arrival rate is maintained, A disk array system comprising: an index estimation unit that outputs the estimation result and a service unit that displays the estimation result. 3. A host computer, a disk controller, and a plurality of disk drives, which record data in a distributed manner on the plurality of disk drives, and when one or more disk drives fail, hold the failed disk. By restoring the data that was restored based on the data in the normal disk drive and writing it to the repaired / replaced disk drive or spare disk drive at the copy speed specified by the operator. In a disk array system that performs recovery processing against the failure while continuing online processing, the arrival rate of input / output requests from the host computer, the disk recovery time until the completion of copy processing, the copy speed, and the disk recovery process Measure the average I / O request arrival rate,
Monitor means for recording them as a monitor result, and based on the monitor result, the relationship between the copy speed and the disk recovery time when the current input / output request arrival rate is assumed to be maintained is found, and the disk Based on the recovery time, the failure rate of one disk drive, and the repair replacement rate, calculate the average time MTTDL until data cannot be restored due to the failure of multiple disk devices, and output it as an estimation result. A disk array system, comprising: an index estimating unit that performs the above-mentioned estimation; and a service unit that displays the estimation result as an index for designating the copy speed and allows an operator to designate the copy speed. 4. A host computer, a disk controller, and a plurality of disk drives, which record data in a distributed manner on the plurality of disk drives, and retain the failed disk when one or more disk drives fail. By executing the copy process that restores the existing data from the data in the normal disk drive and writes it to the repaired / replaced disk drive or spare disk drive at the copy speed specified by the operator. , In a disk array system that performs recovery processing against the failure while continuing online processing, the arrival rate of I / O requests from the host computer and the average until the completion of processing of I / O requests from the host computer during disk recovery processing Response time, copy speed, and flatness during disk recovery processing Monitoring means for measuring the I / O request arrival rate and recording them as a monitoring result, and a copy speed and an average response assuming that the I / O request arrival rate at the present time is maintained based on the monitoring result. It is characterized by comprising index estimation means for obtaining a time relationship and outputting it as an estimation result, and service means for displaying the estimation result as an index for copy speed designation and for causing an operator to designate the copy speed. Disk array system. 5. A host computer, a disk controller, and a plurality of disk drives, which record data in a distributed manner on the plurality of disk drives, and retain the failed disk when one or more disk drives fail. By restoring the data that was restored based on the data in the normal disk drive and writing it to the repaired / replaced disk drive or spare disk drive at the copy speed specified by the operator. , In a disk array system that performs recovery processing against the failure while continuing online processing, arrival rate of I / O requests from the host computer, response time until completion of processing of I / O requests from the host computer, and completion of copy processing Disk recovery time, copy speed, disk Monitor means for measuring the average input / output request arrival rate during the old process and recording them as a monitor result, and a case where it is assumed that the current input / output request arrival rate is maintained based on the monitor result. Each copy speed when the relationship between the copy speed and the disk recovery time is obtained, and based on the target response time specified by the operator and the monitoring result, it is assumed that the current I / O request arrival rate is maintained. Response time is estimated. If the response time is smaller than the target response time, the performance is maintained, and if the response time is larger, the performance is deteriorated. Then, the disk recovery time and the failure rate of one disk drive are calculated. Based on the repair replacement rate and the status classification, the average performance that functions in the performance maintaining status until the data cannot be restored due to the failure of multiple disk drives. An index estimating means for calculating the holding time and outputting it as an estimation result, and a service means for displaying the estimation result as an index for specifying a copy speed and for causing an operator to specify the copy speed are provided. Disk array system. 6. The disk array system according to any one of claims 1 to 5, wherein the monitor means checks the copy processing execution status and records it in the monitor result, and the index estimation means, while the copy processing is being executed, The remaining time required for disk recovery is estimated based on the monitor result and recorded in the estimated result. The service means displays the estimated result as a copy speed change instruction index during the copying process and copies it to the operator. A disk array system comprising a copy processing means for changing a copy speed when a copy processing operator gives a copy speed changing instruction. 7. The disk array system according to claim 1, wherein the monitor means checks the copy processing execution status and records it in the monitor result, and the service means receives the target response time from the operator. Further, when the average response time is shorter than the target response time based on the monitor result and the target response time during the copy process execution, the copy speed is changed to a larger value or the copy execution frequency is increased. If the average response time is longer than the target response time, the copy speed is changed to a smaller value or the copy execution frequency is changed to a lower value, and the copy condition is changed by the copy adjusting means. A disk array system characterized by comprising a copy processing means for changing a copy condition and performing a copy process. 8. The disk array system according to claim 7, wherein when a spare disk drive is provided and a target response speed and a copy speed at the start of copying are given in advance, without waiting for an instruction from an operator. , A disk array system in which each means starts an operation for copy processing. 9. A host computer, a disk controller, and a plurality of disk drives, which record data in a distributed manner on the plurality of disk drives, and when one or more disk drives fail, hold the failed disk. Restore the previously stored data based on the data in the normal disk drive, and write the data to the repaired / replaced disk drive or spare disk drive at the specified copy speed to perform online processing. In a disk array system that continuously performs recovery processing for the failure, the arrival rate of I / O requests from the host computer, the response time until the completion of processing of the I / O requests from the host computer, and the disk recovery until the completion of copy processing Time, copy speed, disk recovery process Monitoring means for measuring the uniform input / output request arrival rate and recording them as a monitoring result, and a copy speed and a disk assuming that the current input / output request arrival rate is maintained based on the monitoring result. Based on the target response time specified by the operator and the monitoring result, the response time for each copy speed when the I / O request arrival rate is assumed to be maintained at the present time is calculated. If the average response time is smaller than the target response time, the performance is maintained, and if the average response time is larger than the target response time, the performance is deteriorated. Then, the disk recovery time, the failure rate of one disk drive and the replacement rate are repaired. Based on the above and the status classification, the average performance maintenance time that is functioning in the above performance maintenance state until the data cannot be restored due to the failure of multiple disk drives. Calculated and,
An index estimation unit that outputs it as an estimation result, a copy adjustment unit that refers to the estimation result and sets a copy speed that maximizes the average performance maintenance time, and a copy process that uses the copy speed set by the copy adjustment unit. And a copy processing means for starting the disk array system. 10. The disk array system according to claim 9, wherein the monitoring means measures the response time, the number of tracks for which copying is completed and the elapsed time from the start of copying, and sets the result as a monitoring result. The index estimation means re-estimates the average performance maintenance time for each copy speed based on the target response time specified by the operator and the monitoring result during the copying process, and sets it as the estimation result. The copy adjusting unit refers to the estimation result at an appropriate cycle and changes the copy speed so that the average performance maintaining time becomes maximum, and the copying unit copies the copy speed when the copy adjusting unit changes the copy speed. A disk array system characterized by performing copy processing after changing to speed. 11. The disk array system according to claim 9 or 10, wherein each unit has a spare disk drive and a target response speed is given in advance, without waiting for an instruction from an operator. The disk array system, wherein the disk array system starts an operation for a copy process.