System and method for controlling power to resources based on historical utilization data

• US 8,041,967 B2
• Filed: 02/15/2005
• Issued: 10/18/2011
• Est. Priority Date: 02/15/2005
• Status: Active Grant

• Current Assignee: MOSAID Technologies Inc. (f/k/a Conversant Intellectual Property Management)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Vaden, Tom L., Belady, Christian L.
• Primary Examiner(s): REHMAN, MOHAMMED H
• Application Number:
  US11/058,554
• Publication Number:
  US 20060184287A1
• Time in Patent Office: 2,436 Days
• Field of Search: None
• US Class Current: 713/320
• CPC Class Codes: G06F 1/3203 Power management, i.e. event-based initiation of a power-saving mode
• CPC Class Codes: G06F 1/325 Power saving in peripheral device
• CPC Class Codes: G06F 1/3275 Power saving in memory, e.g. RAM, cache
• CPC Class Codes: G06F 1/3287 by switching off individual functional units in the computer system
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A method comprises collecting utilization data for a resource, and predicting by a power management system, based on the collected utilization data, future utilization of the resource. The method further comprises controlling, by the power management system, power to the resource, based at least in part on the predicted future utilization of the resource. In one embodiment, the utilization data is collected for a plurality of resources that are operable to perform tasks, and the method further comprises determining, by the power management system, how many of the resources are needed to provide a desired capacity for servicing the predicted future utilization of the resources for performing the tasks. The method further comprises configuring, by the power management system, ones of the resources exceeding the determined number of resources needed to provide the desired capacity in a reduced power-consumption mode.

Claims

1. A method comprising:

collecting utilization data for a resource;

inputting parameters specifying a desired capacity of said resource;

predicting by a power management system, based on said collected utilization data, future utilization of said resource, wherein said parameters further specify that an amount of extra capacity above said predicted future utilization is to vary based on a volatility in said predicted future utilization of said resource; and

controlling, by the power management system, power to said resource, based at least in part on said predicted future utilization of said resource and said input parameters.

2. The method of claim 1 wherein said collecting, predicting, and controlling comprise:

collecting utilization data for a plurality of resources;

predicting by said power management system, based on said collected utilization data, future utilization of said resources; and

controlling, by the power management system, power to said resources, based at least in part on said predicted future utilization of said resources.

3. The method of claim 2 wherein said resources comprise computing resources in a system.

4. The method of claim 2 wherein said resources comprise servers in a data center.

5. The method of claim 2 wherein said resources comprise ones selected from the group consisting of:

processors, memory, cells, servers, input/output (I/O) devices, power supplies, heating devices, cooling devices, switches, and routers.

6. The method of claim 2 wherein said controlling comprises:

said power management system configuring ones of said resources into a power-conserving mode.

7. The method of claim 6 further comprising:

absorbing by fully powered ones of said resources workload of said ones of said resources that are configured into said power-conserving mode.

8. The method of claim 6 wherein said controlling further comprises:

said power management system configuring ones of said resources that are in said power-conserving mode into a full-power mode.

9. The method of claim 8 further comprising:

said power management system beginning said configuring of said ones of said resources that are in said power-conserving mode into said full-power mode sufficiently in advance of a time of predicted utilization of said ones of said resources such that said ones of said resources are in said full-power mode by said time of said predicted utilization thereof.

10. The method of claim 2 wherein said power management system is nested in a hierarchy of power management systems.

11. The method of claim 1 wherein said controlling comprises:

dynamically configuring said resource into a power-conserving mode.

12. The method of claim 11 wherein said controlling further comprises:

selecting, by said power management system, one of a plurality of different power-conserving modes in which to configure said resource.

13. The method of claim 12 wherein said selecting comprises:

selecting said one of a plurality of different power-conserving modes based at least in part on said predicted future utilization of said resource.

14. The method of claim 12 wherein said selecting comprises:

selecting said one of a plurality of different power-conserving modes based at least in part on a predicted length of time that said resource will not be utilized.

15. The method of claim 11 wherein said plurality of different power-conserving modes comprise ones selected from the group consisting of:

hard power down, soft power down, standby, sleep, hibernate, and power-conserving P-state level.

16. A method comprising:

collecting utilization data for resources that are operable to perform tasks;

inputting parameters specifying a desired capacity of said resources performing said tasks;

predicting by a power management system, based on said collected utilization data, future utilization of said resources for performing said tasks, wherein said input parameters further specify that an amount of extra capacity above said predicted future utilization is to vary based on a volatility in said predicted future utilization of said resources;

determining, by said power management system, how many of said resources are needed to provide the desired capacity and the extra capacity for servicing the predicted future utilization of the resources for performing said tasks; and

configuring, by said power management system, ones of the resources exceeding the determined number of resources needed to provide the desired capacity and the extra capacity in a reduced power-consumption mode.

17. The method of claim 16 further comprising:

assigning, by the power management system, said tasks to the determined number of resources needed to provide the desired capacity.

18. The method of claim 16 wherein the configuring is performed dynamically by said power management system.

19. The method of claim 16 further comprising:

upon predicting by the power management system that ones of said resources configured in said reduced power-consumption mode are needed to provide said desired capacity for servicing a predicted future utilization, said power management system configuring ones of the resources that are in a reduced power-consumption mode into a full power-consumption mode.

20. The method of claim 19 further comprising:

said power management system beginning said configuring of said ones of said resources that are in said reduced power-consumption mode into said full power-consumption mode sufficiently in advance of a time of predicted utilization of said ones of said resources such that said ones of said resources are in said full power-consumption mode by said time of said predicted utilization thereof.

21. The method of claim 16 wherein said resources comprise servers in a data center.

22. The method of claim 16 further comprising:

absorbing by fully powered ones of said resources workload of said ones of said resources that are configured into said reduced power-consumption mode.

23. The method of claim 16 wherein said configuring further comprises:

selecting, by said power management system, one of a plurality of different reduced power-consumption modes in which to configure said ones of the resources exceeding the determined number of resources needed to provide the desired capacity.

24. The method of claim 23 wherein said selecting comprises:

selecting said one of a plurality of different reduced power-consumption modes based at least in part on said predicted future utilization of said resources.

25. The method of claim 16 wherein said plurality of different reduced power-consumption modes comprise ones selected from the group consisting of:

hard power down, soft power down, standby, sleep, hibernate, and power-conserving P-state level.

26. The method of claim 16 further comprising:

receiving input to said power management system for defining at least the desired capacity.

27. The method of claim 16 further comprising:

configuring, by said power management system, at least one related resource into a reduced power-consumption mode based at least in part on configuring said ones of the resources exceeding the determined number of resources needed to provide the desired capacity into said reduced power-consumption mode.

28. The method of claim 27 wherein said at least one related resource comprises power supplies for supplying power at least to the ones of the resources that are configured into said reduced power-consumption mode.

29. The method of claim 28 wherein said configuring said at least one related resource into a reduced power-consumption mode comprises:

powering down ones of said power supplies to maintain remaining ones of said power supplies at an optimum load.

30. The method of claim 27 wherein said at least one related resource comprises a cooling system for cooling the ones of the resources that are configured into said reduced power-consumption mode.

31. The method of claim 16 wherein said power management system comprises one of a plurality of nested power management systems.

32. A system comprising:

a power-consuming resource;

a console for inputting parameters specifying a desired capacity of the power-consuming resource;

utilization monitor for collecting utilization data corresponding to utilization of the power-consuming resource; and

power management logic for predicting, based on the collected utilization data, future utilization of said power-consuming resource, and for controlling power to said power-consuming resource based at least in part on the predicted future utilization thereof and in part on said input parameters, wherein said inputted parameters further specify that an amount of extra capacity above said predicted future utilization is to vary based on a volatility in said predicted future utilization of the power consuming resource, and wherein said controlling power is also based at least in part on said amount of extra capacity.

33. The system of claim 32 comprising:

a plurality of said power-consuming resources;

said utilization monitor for collecting utilization data corresponding to utilization of the power-consuming resources; and

said power management logic for predicting, based on the collected utilization data, future utilization of said power-consuming resources, and for controlling power to said power-consuming resources based at least in part on the predicted future utilization thereof.

34. The system of claim 33 wherein said resources comprise servers in a data center.

35. The system of claim 33 wherein said resources comprise ones selected from the group consisting of:

processors, memory, cells, servers, input/output (I/O) devices, power supplies, heating devices, cooling devices, switches, and routers.

36. The system of claim 33 wherein said power management logic for controlling power to said power-consuming resources comprises:

logic for configuring ones of said resources into a power-conserving mode.

37. The system of claim 36 further comprising:

fully powered ones of said resources for absorbing workload of said ones of said resources that are configured into said power-conserving mode.

38. The system of claim 36 wherein said power management logic for controlling power to said power-consuming resources further comprises:

logic for configuring ones of said resources that are in said power-conserving mode into a full-power mode.

39. The system of claim 38 wherein said power management logic for configuring ones of said resources that are in said power-conserving mode into a full-power mode further comprises:

logic for beginning said configuring of said ones of said resources that are in said power-conserving mode into said full-power mode sufficiently in advance of a time of predicted utilization of said ones of said resources such that said ones of said resources are in said full-power mode by said time of said predicted utilization thereof.

40. The system of claim 32 wherein said power management logic controlling power to said power-consuming resources further comprises:

logic for selecting one of a plurality of different power-conserving modes in which to configure said resource based at least in part on said predicted future utilization of said resource.

41. The system of claim 32 wherein said plurality of different power-conserving modes comprise ones selected from the group consisting of:

hard power down, soft power down, standby, sleep, hibernate, and power-conserving P-state level.

42. The system of claim 32 further comprising:

a second power-consuming resource nested in a hierarchy of power consuming resources with said power-consuming resource so that said power-consuming resource is contained within said second power-consuming resource;

a second utilization monitor for collecting utilization data corresponding to utilization of the second power-consuming resource; and

second power management logic for predicting future utilization of said second power-consuming resource, and for controlling power to said second power-consuming resource.

43. A system comprising:

power-consuming resources for performing tasks;

means for inputting parameters specifying a desired capacity of the resources for performing the tasks;

means for collecting utilization data corresponding to utilization of the resources for performing the tasks;

means for predicting, based on the collected utilization data, future utilization of said resources; and

means for controlling a power-consumption mode of the resources based at least in part on the predicted future utilization of said resources and in part on said input parameters, wherein said parameters further specify that an amount of extra capacity above said predicted future utilization is to vary based on a volatility in said predicted future utilization of said resource, said power consumption mode being further based on said amount of extra capacity.

44. The system of claim 43 wherein the controlling means comprises:

means for determining how many of said resources are needed to provide a desired capacity for servicing the predicted future utilization; and

means for configuring ones of the resources exceeding the determined number of resources needed to provide the desired capacity in a reduced power-consumption mode.

45. The system of claim 43 wherein said power-consuming resources comprise servers in a data center.

46. A data center comprising:

a plurality of servers;

a console for inputting parameters specifying a desired capacity of the plurality of servers;

utilization monitor for collecting utilization data corresponding to utilization of the servers; and

power management logic for predicting, based on the collected utilization data, future utilization of said servers, wherein said parameters further specify that an amount of extra capacity above said predicted future utilization is to vary based on a volatility in said predicted future utilization of said servers, said power management logic dynamically configuring ones of said servers into appropriate power-consuming modes based at least in part on the predicted future utilization thereof and in part on said input parameters.

47. The data center of claim 46 wherein said power management logic for dynamically configuring ones of said servers into appropriate power-consuming modes comprises:

logic for determining how many of said servers are needed to provide a desired capacity for servicing the predicted future utilization; and

logic for configuring ones of the servers exceeding the determined number of servers needed to provide the desired capacity in a reduced power-consumption mode.

48. The data center of claim 46 wherein said plurality of different reduced power-consumption modes comprise ones selected from the group consisting of:

hard power down, soft power down, standby, sleep, hibernate, and power-conserving P-state level.