Selecting a priority queue from which to process an input/output (I/O) request by training a machine learning module

• US 10,558,596 B2
• Filed: 08/01/2018
• Issued: 02/11/2020
• Est. Priority Date: 05/18/2018
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Nielsen, Karl A., Hardy, Clint A., Gupta, Lokesh M., Borlick, Matthew G.
• Primary Examiner(s): Snyder, Steven G
• Application Number:
  US16/052,518
• Publication Number:
  US 20190354497A1
• Time in Patent Office: 559 Days
• Field of Search: None
• CPC Class Codes: G06F 13/1673 using buffers
• CPC Class Codes: G06F 13/225 with priority control
• CPC Class Codes: G06F 3/0611 in relation to response time
• CPC Class Codes: G06F 3/0656 Data buffering arrangements
• CPC Class Codes: G06F 3/0659 Command handling arrangements, e.g. command buffers, queues, command scheduling
• CPC Class Codes: G06F 3/0689 Disk arrays, e.g. RAID, JBOD
• CPC Class Codes: G06F 5/06 for changing the speed of data flow, i.e. speed regularising or timing, e.g. delay lines, FIFO buffers; over- or underrun control therefor G06F7/78 takes precedence
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent

Abstract

Provided are a computer program product, system, and method for using at least one machine learning module to select a priority queue from which to process an Input/Output (I/O) request. Input I/O statistics are provided on processing of I/O requests at the queues to at least one machine learning module. Output is received from the at least one machine learning module for each of the queues. The output for each queue indicates a likelihood that selection of an I/O request from the queue will maintain desired response time ratios between the queues. The received output for each of the queues is used to select a queue of the queues. An I/O request from the selected queue is processed.

Claims

1. A computer program product for selecting one of a plurality queues having Input/Output (I/O) requests for a storage to process, comprising a computer readable storage medium having computer readable program code embodied therein that when executed performs operations, the operations comprising:

providing a machine learning module that receives as input I/O statistics for the queues based on I/O activity at the queues and produces output for each of the queues, wherein the output for each queue indicates a likelihood that selection of an I/O request from the queue will maintain expected response time ratios between the queues, wherein outputs for the queues from the machine learning module are included in a task control block;

using the outputs to select a queue of the queues;

servicing the I/O request from the selected queue;

after servicing the I/O request, determining adjusted outputs from the outputs for the queues indicated in the task control block;

retraining the machine learning module with the input I/O statistics to produce the adjusted outputs for the queues; and

using the retrained machine learning module to select one of the queues from which to process an I/O request.

2. The computer program product of claim 1, wherein the determining the adjusted outputs and retraining the machine learning module are performed after processing at least one I/O request from one of the queues.

3. The computer program product of claim 1, wherein the adjusted outputs are determined for less than all of the queues, wherein the retraining the machine learning module comprises retraining the machine learning module with the input I/O statistics to produce a recent output from the machine learning module for each of at least one queue not having adjusted output determined.

4. The computer program product of claim 1, wherein the determining an adjusted output for one queue comprises calculating the adjusted output as a function of a recent output from the machine learning module for the queue, a desired ratio of a desired response time of the queue and a desired response time of another queue, and a measured ratio of a measured response time for the queue and a measured response time of the another queue.

5. The computer program product of claim 1, wherein the plurality of queues includes a low priority queue, a medium priority queue, and a high priority queue, wherein the medium priority queue stores higher priority I/O requests than the low priority queue, wherein an adjusted output of the high priority queue is determined using a desired high priority ratio of a desired response time for the high priority queue to a desired response time for the medium priority queue and wherein an adjusted output of the medium priority queue is calculated using a desired medium priority ratio of a desired response time for the medium priority queue to a desired response time for the low priority queue.

6. The computer program product of claim 5, wherein the determining the adjusted outputs for the high priority queue and the medium priority queue comprises:

calculating the adjusted output for the high priority queue as a recent output for the high priority queue plus (a constant times (the desired high priority ratio minus (a measured ratio of an average of measured response times for the high priority queue and an average of measured response times for the medium priority queue)); and

calculating the adjusted output for the medium priority queue as a recent output for the medium priority queue plus (a constant times (the desired medium priority ratio minus (a measured ratio of an average of measured response times for the medium priority queue and an average of measured response times for the low priority queue))).

7. The computer program product of claim 5, wherein the machine learning module is retrained using adjusted outputs for the high priority queue and the medium priority queue and a recent output for the low priority queue calculated by the machine learning module.

8. The computer program product of claim 1, wherein the operations further comprise:

determining whether any of the queues have an average response time greater than an acceptable maximum average response time; and

adjusting at least one of the adjusted outputs and recent outputs from the machine learning module used to retrain the machine learning module to increase an adjusted or recent output of a queue having an average response time greater than the acceptable maximum average response time relative to any queues having average response time less than the acceptable maximum average response time.

9. A system for selecting one of a plurality queues having Input/Output (I/O) requests for a storage to process, comprising:

a processor; and

a computer readable storage medium having computer readable program code embodied therein that when executed performs operations, the operations comprising;

providing at least one machine learning module that receives as input I/O statistics for the queues based on I/O activity at the queues and produces output for each of the queues, wherein the output for each queue indicates a likelihood that selection of an I/O request from the queue will maintain expected response time ratios between the queues, wherein outputs for the queues from the machine learning module are included in a task control block;

using the outputs to select a queue of the queues;

servicing the I/O request from the selected queue;

after servicing the I/O request, determining adjusted outputs from the outputs for the queues indicated in the task control block;

retraining the machine learning module with the input I/O statistics to produce the adjusted outputs for the queues; and

using the retrained machine learning module to select one of the queues from which to process an I/O request.

10. The system of claim 9, wherein the plurality of queues includes a low priority queue, a medium priority queue, and a high priority queue, wherein the medium priority queue stores higher priority I/O requests than the low priority queue, wherein an adjusted output of the high priority queue is determined using a desired high priority ratio of a desired response time for the high priority queue to a desired response time for the medium priority queue and wherein an adjusted output of the medium priority queue is calculated using a desired medium priority ratio of a desired response time for the medium priority queue to a desired response time for the low priority queue.

11. The system of claim 10, wherein the operations further comprise:

determining whether any of the queues have an average response time greater than an acceptable maximum average response time; and

adjusting at least one of the adjusted outputs and recent outputs from the machine learning module used to retrain the machine learning module to increase an adjusted or recent output of a queue having an average response time greater than the acceptable maximum average response time relative to any queues having average response time less than the acceptable maximum average response time.

15. The system of claim 9, wherein the determining the adjusted outputs and retraining the machine learning module are performed after processing at least one I/O request from one of the queues.

16. The system of claim 9, wherein the adjusted outputs are determined for less than all of the queues, wherein retraining the machine learning module comprises retraining the machine learning module with the input I/O statistics to produce a recent output from the machine learning module for each of at least one queue not having adjusted output determined.

17. The system of claim 9, wherein the determining an adjusted output for one queue comprises calculating the adjusted output as a function of a recent output from the machine learning module for the queue, a desired ratio of a desired response time of the queue and a desired response time of another queue, and a measured ratio of a measured response time for the queue and a measured response time of the another queue.

18. The system of claim 10, wherein the determining the adjusted outputs for the high priority queue and the medium priority queue comprises:

calculating the adjusted output for the high priority queue as a recent output for the high priority queue plus (a constant times (the desired high priority ratio minus (a measured ratio of an average of measured response times for the high priority queue and an average of measured response times for the medium priority queue)); and

calculating the adjusted output for the medium priority queue as a recent output for the medium priority queue plus (a constant times (the desired medium priority ratio minus (a measured ratio of an average of measured response times for the medium priority queue and an average of measured response times for the low priority queue))).

19. The system of claim 10, wherein the machine learning module is retrained using adjusted outputs for the high priority queue and the medium priority queue and a recent output for the low priority queue calculated by the at least one machine learning module.

12. A method for selecting one of a plurality queues having Input/Output (I/O) requests for a storage to process, comprising:

providing at least one machine learning module that receives as input I/O statistics for the queues based on I/O activity at the queues and produces output for each of the queues, wherein the output for each queue indicates a likelihood that selection of an I/O request from the queue will maintain expected response time ratios between the queues, wherein outputs for the queues from the machine learning module are included in a task control block;

using the outputs to select a queue of the queues;

servicing the I/O request from the selected queue;

after servicing the I/O request, determining adjusted outputs from the outputs for the queues indicated in the task control block;

retraining the machine learning module with the input I/O statistics to produce the adjusted outputs for the queues; and

using the retrained machine learning module to select one of the queues from which to process an I/O request.

13. The method of claim 12, wherein the plurality of queues includes a low priority queue, a medium priority queue, and a high priority queue, wherein the medium priority queue stores higher priority I/O requests than the low priority queue, wherein an adjusted output of the high priority queue is determined using a desired high priority ratio of a desired response time for the high priority queue to a desired response time for the medium priority queue and wherein an adjusted output of the medium priority queue is calculated using a desired medium priority ratio of a desired response time for the medium priority queue to a desired response time for the low priority queue.

14. The method of claim 12;

further comprising;

determining whether any of the queues have an average response time greater than an acceptable maximum average response time; and

adjusting at least one of the adjusted outputs and recent outputs from the machine learning module used to retrain the machine learning module to increase an adjusted or recent output of a queue having an average response time greater than the acceptable maximum average response time relative to any queues having average response time less than the acceptable maximum average response time.

20. The method of claim 12, wherein the determining the adjusted outputs and retraining the machine learning module are performed after processing at least one I/O request from one of the queues.

21. The method of claim 12, wherein the adjusted outputs are determined for less than all of the queues, wherein the retraining the machine learning module comprises retraining the machine learning module with the input I/O statistics to produce a recent output from the machine learning module for each of at least one queue not having adjusted output determined.

22. The method of claim 12, wherein the determining an adjusted output for one queue comprises calculating the adjusted output as a function of a recent output from the machine learning module for the queue, a desired ratio of a desired response time of the queue and a desired response time of another queue, and a measured ratio of a measured response time for the queue and a measured response time of the another queue.

23. The method of claim 13, wherein the determining the adjusted outputs for the high priority queue and the medium priority queue comprises:

calculating the adjusted output for the high priority queue as a recent output for the high priority queue plus (a constant times (the desired high priority ratio minus (a measured ratio of an average of measured response times for the high priority queue and an average of measured response times for the medium priority queue))); and

calculating the adjusted output for the medium priority queue as a recent output for the medium priority queue plus (a constant times (the desired medium priority ratio minus (a measured ratio of an average of measured response times for the medium priority queue and an average of measured response times for the low priority queue)).

24. The method of claim 13, wherein the machine learning module is retrained using adjusted outputs for the high priority queue and the medium priority queue and a recent output for the low priority queue calculated by the machine learning module.