System overhead-based automatic adjusting of number of running processors within a system

• US 9,292,348 B2
• Filed: 07/16/2013
• Issued: 03/22/2016
• Est. Priority Date: 07/16/2013
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Carollo, Mary Ellen, Wade, Brian K., Wilton, Donald P., Adams, Kevin S.
• Primary Examiner(s): Wu, Qing
• Application Number:
  US13/942,901
• Publication Number:
  US 20150026697A1
• Time in Patent Office: 980 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 2209/5022 Workload threshold
• CPC Class Codes: G06F 9/5027 the resource being a machine, e.g. CPUs, Servers, Terminals
• CPC Class Codes: G06F 9/505 considering the load
• CPC Class Codes: G06F 9/5077 Logical partitioning of resources; Management or configuration of virtualized resources specific details on emulation or internal functioning of virtual machines G06F9/455
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system

Abstract

Data processing system efficiency is improved by automatically determining whether to adjust for a next time interval a number N of processors running within the system for processing a workload. The automatically determining includes obtaining a measure of operating system overhead by evaluating one or more characteristics of processor time of the N processors consumed within the system for a time interval, and obtaining a measure of system utilization of the N processors running within the system for processing the workload for the time interval. The automatically determining further includes automatically ascertaining whether to adjust the number N of processors running within the system for the next time interval to improve system efficiency using the obtained measure of operating system overhead and the obtained measure of system utilization of the N processors.

Claims

1. A computer program product comprising:

a non-transitory computer-readable storage medium readable by a processor and storing executable instructions for execution by the processor for performing a method comprising;

automatically determining whether to adjust a number N of processors running within a system for processing a workload, the system comprising an operating system, and the automatically determining comprising;

obtaining a measure of operating system overhead by evaluating one or more characteristics of processor time of the N processors consumed within the system for a time interval;

obtaining a measure of system utilization of the N processors running within the system for processing the workload for the time interval; and

automatically ascertaining whether to adjust the number N of processors running within the system for a next time interval to improve system efficiency using, at least in part, the obtained measure of operating system overhead and the obtained measure of system utilization of the N processors.

2. The computer program product of claim 1, wherein the measure of operating system overhead comprises a ratio of total processor time consumed for the time interval (T) over the sum of processor time consumed for the time interval by applications (V) running within the system.

3. The computer program product of claim 2, wherein the method further comprises periodically repeating the automatically determining to, in part, obtain a sample history of the measure of operating system overhead and a sample history of the measure of system utilization of the N processors for different time intervals, and wherein the automatically ascertaining comprises utilizing the sample histories of the measure of operating system overhead and the measure of system utilization of the N processors to project an operating system overhead ceiling (R′

) for the next time interval and a system utilization ceiling (U′

) for the next time interval, and the automatically ascertaining comprises utilizing the projected operating system overhead ceiling (R′

) and projected system utilization ceiling (U′

) in determining whether to adjust the number N of processors running within the system for the next time interval to improve system efficiency.

4. The computer program product of claim 3, further comprising, after repeating the automatically determining x+1 times, discarding an oldest measure of operating system overhead from the sample history of the measure of operating system overhead and discarding an oldest measure of system utilization of the N processors from the sample history of the measure of system utilization of the N processors.

5. The computer program product of claim 4, wherein the automatically ascertaining comprises automatically determining a participating amount (A) of total processor capacity needed to participate in running the workload over the next time interval.

6. The computer program product of claim 5, wherein the automatically ascertaining further comprises comparing the projected operating system overhead ceiling (R′

) to at least one system overhead threshold and determining, based at least in part thereon, whether to adjust the number N of processors miming within the system for the next time interval to improve system efficiency.

7. The computer program product of claim 5, wherein the automatically ascertaining further comprises determining whether the projected operating system overhead ceiling (R′

) is below a low system overhead threshold, and if so, setting the participating amount (A) of total processor capacity to full processor capacity for the next time interval.

8. The computer program product of claim 5, wherein the automatically determining further comprises determining whether the projected operating system overhead ceiling (R′

) is above an upper system overhead threshold, and if so, setting the participating amount (A) of total processor capacity to the projected operating system utilization ceiling (U′

) plus a set margin.

9. The computer program product of claim 5, wherein the automatically determining further comprises determining whether the projected operating system overhead ceiling (R′

) is between a low system overhead threshold and a high system overhead threshold, and if so, setting the participating amount (A) of total processor capacity for the next time interval to a value between full processor capacity, and the projected system utilization ceiling (U′

) plus a set margin.

10. The computer program product of claim 5, further comprising automatically placing one or more processors of the N processors running within the system into a sleep state responsive to determining that the participating amount (A) of total processor capacity needed to participate in running the workload over the next time interval is less than the number N of processors currently running within the system.

11. A computer system comprising:

a memory; and

a processor in communications with the memory, wherein the computer system configured to perform a method comprising;

automatically determining whether to adjust a number N of processors running within the computer system for processing a workload, the computer system comprising an operating system and the automatically determining comprising;

obtaining a measure of operating system overhead by evaluating one or more characteristics of processor time of the N processors consumed within the system for a time interval;

obtaining a measure of system utilization of the N processors running within the system for processing the workload for the time interval; and

automatically ascertaining whether to adjust the number N of processors running within the system for a next time interval to improve system efficiency using, at least in part, the obtained measure of operating system overhead and the obtained measure of system utilization of the N processors.

12. The computer system of claim 11, wherein the measure of operating system overhead comprises a ratio of total processor time consumed for the time interval (T) over the sum of processor time consumed for the time interval by applications (V) running within the system.

13. The computer system of claim 12, wherein the method further comprises periodically repeating the automatically determining to, in part, obtain a sample history of the measure of operating system overhead and a sample history of the measure of system utilization of the N processors for different time intervals, and wherein the automatically ascertaining comprises utilizing the sample histories of the measure of operating system overhead and the measure of system utilization of the N processors to project an operating system overhead ceiling (R′

) for the next time interval and a system utilization ceiling (U′

) for the next time interval, and the automatically ascertaining comprises utilizing the projected operating system overhead ceiling (R′

) and projected system utilization ceiling (U′

) in determining whether to adjust the number N of processors running within the system for the next time interval to improve system efficiency.

14. The computer system of claim 13, further comprising, after repeating the automatically determining x+1 times, discarding an oldest measure of operating system overhead from the sample history of the measure of operating system overhead and discarding an oldest measure of system utilization of the N processors from the sample history of the measure of system utilization of the N processors.

15. The computer system of claim 14, wherein the automatically ascertaining further comprises comparing the projected operating system overhead ceiling (R′

) to at least one system overhead threshold and determining, based at least in part thereon, whether to adjust the number N of processors running within the system for the next time interval to improve system efficiency.

16. A method comprising:

automatically determining whether to adjust a number N of processors running within a system for processing a workload, the system comprising an operating system and the automatically determining comprising;

obtaining a measure of operating system overhead by evaluating one or more characteristics of processor time of the N processors consumed within the system for a time interval;

obtaining a measure of system utilization of the N processors running within the system for processing the workload for the time interval; and

automatically ascertaining whether to adjust the number N of processors running within the system for a next time interval to improve system efficiency using, at least in part, the obtained measure of operating system overhead and the obtained measure of system utilization of the N processors.

17. The method of claim 16, wherein the measure of operating system overhead comprises a ratio of total processor time consumed for the time interval (T) over the sum of processor time consumed for the time interval by applications (V) running within the system.

18. The method of claim 17, further comprising periodically repeating the automatically determining to, in part, obtain a sample history of the measure of operating system overhead and a sample history of the measure of system utilization of the N processors for different time intervals, and wherein the automatically ascertaining comprises utilizing the sample histories of the measure of operating system overhead and the measure of system utilization of the N processors to project an operating system overhead ceiling (R′

) for the next time interval and a system utilization ceiling (U′

) for the next time interval, and the automatically ascertaining comprises utilizing the projected operating system overhead ceiling (R′

) and projected system utilization ceiling (U′

) in determining whether to adjust the number N of processors running within the system for the next time interval to improve system efficiency.

19. The method of claim 18, further comprising, after repeating the automatically determining x+1 times, discarding an oldest measure of operating system overhead from the sample history of the measure of operating system overhead and discarding an oldest measure of system utilization of the N processors from the sample history of the measure of system utilization of the N processors.

20. The method of claim 19, wherein the automatically ascertaining further comprises comparing the projected operating system overhead ceiling (R′

) to at least one system overhead threshold and determining, based at least in part thereon, whether to adjust the number N of processors running within the system for the next time interval to improve system efficiency.