Scheduling storage unit maintenance tasks in a dispersed storage network

• US 10,055,170 B2
• Filed: 09/12/2016
• Issued: 08/21/2018
• Est. Priority Date: 04/30/2015
• Status: Active Grant

• Current Assignee: Pure Storage, Inc.
• Original Assignee: International Business Machines Corporation
• Inventors: Resch, Jason K., Wozniak, Ethan S.
• Primary Examiner(s): Bryan, Jason B
• Application Number:
  US15/262,654
• Publication Number:
  US 20160378405A1
• Time in Patent Office: 708 Days
• CPC Class Codes: G06F 11/008 Reliability or availability analysis
• CPC Class Codes: G06F 11/1092 Rebuilding, e.g. when physically replacing a failing disk
• CPC Class Codes: G06F 2211/1028 Distributed, i.e. distributed RAID systems with parity
• CPC Class Codes: G06F 3/0604 Improving or facilitating administration, e.g. storage management
• CPC Class Codes: G06F 3/0619 in relation to data integrity, e.g. data losses, bit errors
• CPC Class Codes: G06F 3/064 Management of blocks
• CPC Class Codes: G06F 3/0653 Monitoring storage devices or systems
• CPC Class Codes: G06F 3/0659 Command handling arrangements, e.g. command buffers, queues, command scheduling
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]

Abstract

A method for execution by a dispersed storage and task (DST) execution unit includes generating low-load prediction data, which includes selecting a time period corresponding to a predicted low-load, based on a plurality of historical load samplings. Maintenance task scheduling data is generated based on the low-load prediction data. Generating the maintenance task scheduling data includes assigning a maintenance task to a scheduled time that is within the time period corresponding to the predicted low-load. The maintenance task is executed at the scheduled time.

Claims

1. A method for execution by a dispersed storage and task (DST) execution unit that includes a processor, the method comprises:

generating low-load prediction data based on a plurality of historical load samplings, wherein generating the low-load prediction data includes selecting a time period corresponding to a predicted low-load;

generating maintenance task scheduling data based on the low-load prediction data, wherein generating the maintenance task scheduling data includes assigning a maintenance task to a scheduled time that is within the time period corresponding to the predicted low-load;

generating current load data by comparing at least one load reading taken at a time before the time period to a low-load threshold;

executing the maintenance task at the scheduled time when the at least one load reading compares unfavorably to the low-load threshold; and

when the at least one load reading compares favorably to the low-load threshold;

executing the maintenance task at an updated time that is earlier than the scheduled time; and

delaying execution of an access request received via a network by rescheduling the execution of the access request from an original time to a later time in response to the at least one load reading comparing favorably to the low-load threshold, wherein the later time is selected based on the updated time and an expected execution duration of the maintenance task.

2. The method of claim 1, wherein the plurality of historical load samplings include load readings taken from at least one of:

a central processing unit (CPU) of the DST execution unit or a memory device of the DST execution unit.

3. The method of claim 1, further comprising:

updating historical load data that includes the plurality of historical load samplings in response to a predetermined time interval elapsing by adding at least one current load reading to the plurality of historical load samplings.

4. The method of claim 1, wherein the generating the low-load prediction data includes determining a cyclical load pattern based on the historical load samplings, and wherein selecting the time period is based on a minimum of the cyclical load pattern.

5. The method of claim 4, wherein the low-load prediction data includes a plurality of time periods corresponding to a plurality of predicted low-loads, wherein the plurality of time periods are based on a plurality of recurring minimums of the cyclical load pattern, wherein the plurality of time periods are separated by a fixed interval corresponding to a frequency of the cyclical load pattern, and wherein generating the maintenance task scheduling data includes assigning a recurring maintenance task to the plurality of time periods.

6. The method of claim 1, wherein the maintenance task includes scanning a namespace of the DST execution unit.

7. The method of claim 1, wherein the maintenance task includes rebuilding missing data slices stored in the DST execution unit.

8. The method of claim 1, further comprising:

generating current load data by comparing at least one load reading taken during the time period corresponding to the predicted low-load to the low-load prediction data;

wherein the maintenance task is executed at the scheduled time when the at least one load reading compares favorably to the low-load prediction data; and

wherein the maintenance task is rescheduled for execution at a future time that is later than the time period corresponding to the predicted low-load when the at least one load reading compares unfavorably to the low-load prediction data.

9. A processing system of a dispersed storage and task (DST) execution unit comprises:

at least one processor;

a memory that stores operational instructions, that when executed by the at least one processor cause the processing system to;

generate low-load prediction data based on a plurality of historical load samplings, wherein generating the low-load prediction data includes selecting a time period corresponding to a predicted low-load;

generate maintenance task scheduling data based on the low-load prediction data, wherein generating the maintenance task scheduling data includes assigning a maintenance task to a scheduled time that is within the time period corresponding to the predicted low-load;

generate current load data by comparing at least one load reading taken at a time before the time period to a low-load threshold;

execute the maintenance task at the scheduled time when the at least one load reading compares unfavorably to the low-load threshold; and

when the at least one load reading compares favorably to the low-load threshold;

execute the maintenance task at an updated time that is earlier than the scheduled time; and

delay execution of an access request received via a network by rescheduling the execution of the access request from an original time to a later time in response to the at least one load reading comparing favorably to the low-load threshold, wherein the later time is selected based on the updated time and an expected execution duration of the maintenance task.

10. The processing system claim 9, wherein the plurality of historical load samplings include load readings taken from at least one of:

a central processing unit (CPU) of the DST execution unit or a memory device of the DST execution unit.

11. The processing system of claim 9, wherein the operational instructions, when executed by the at least one processor, further cause the processing system to:

update historical load data that includes the plurality of historical load samplings in response to a predetermined time interval elapsing by adding at least one current load reading to the plurality of historical load samplings.

12. The processing system of claim 9, wherein the generating the low-load prediction data includes determining a cyclical load pattern based on the historical load samplings, and wherein selecting the time period is based on a minimum of the cyclical load pattern.

13. The processing system of claim 12, wherein the low-load prediction data includes a plurality of time periods corresponding to a plurality of predicted low-loads, wherein the plurality of time periods are based on a plurality of recurring minimums of the cyclical load pattern, wherein the plurality of time periods are separated by a fixed interval corresponding to a frequency of the cyclical load pattern, and wherein generating the maintenance task scheduling data includes assigning a recurring maintenance task to the plurality of time periods.

14. The processing system of claim 9, wherein the maintenance task includes scanning a namespace of the DST execution unit.

15. The processing system of claim 9, wherein the maintenance task includes rebuilding missing data slices stored in the DST execution unit.

16. The processing system of claim 9, wherein the operational instructions, when executed by the at least one processor, further cause the processing system to:

generate current load data by comparing at least one load reading taken during the time period corresponding to the predicted low-load to the low-load prediction data;

wherein the maintenance task is executed at the scheduled time when the at least one load reading compares favorably to the low-load prediction data; and

wherein the maintenance task is rescheduled for execution at a future time that is later than the time period corresponding to the predicted low-load when the at least one load reading compares unfavorably to the low-load prediction data.

17. A non-transitory computer readable storage medium comprises:

at least one memory section that stores operational instructions that, when executed by a processing system of a dispersed storage network (DSN) that includes a processor and a memory, causes the processing system to;

generate low-load prediction data based on a plurality of historical load samplings, wherein generating the low-load prediction data includes selecting a time period corresponding to a predicted low-load;

generate maintenance task scheduling data based on the low-load prediction data, wherein generating the maintenance task scheduling data includes assigning a maintenance task to a scheduled time that is within the time period corresponding to the predicted low-load;

generate current load data by comparing at least one load reading taken at a time before the time period to a low-load threshold;

execute the maintenance task at the scheduled time when the at least one load reading compares unfavorably to the low-load threshold; and

when the at least one load reading compares favorably to the low-load threshold;

execute the maintenance task at an updated time that is earlier than the scheduled time; and

delay execution of an access request received via a network by rescheduling the execution of the access request from an original time to a later time in response to the at least one load reading comparing favorably to the low-load threshold, wherein the later time is selected based on the updated time and an expected execution duration of the maintenance task.