Power consumption management among compute nodes

• US 20070067657A1
• Filed: 09/22/2005
• Published: 03/22/2007
• Est. Priority Date: 09/22/2005
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Ranganathan, Parthasarathy, Shaver, Charles, Leech, Phillip, Alzien, Khaldoun
• Granted Patent: US 7,647,516 B2
• Time in Patent Office: Days
• US Class Current: 713/320
• CPC Class Codes: G06F 1/206 comprising thermal management
• CPC Class Codes: G06F 1/3203 Power management, i.e. event-based initiation of a power-saving mode
• CPC Class Codes: G06F 1/324 by lowering clock frequency
• CPC Class Codes: G06F 1/3268 Power saving in hard disk drive
• CPC Class Codes: G06F 1/3287 by switching off individual functional units in the computer system
• CPC Class Codes: G06F 1/3296 by lowering the supply or operating voltage
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management
• CPC Class Codes: Y02D 30/50 in wire-line communication networks, e.g. low power modes or reduced link rate

Abstract

In a method for managing power consumption among compute nodes having respective power components, an increase in the power utilization of a first compute node of the compute nodes may be detected. In response to a detected increase, a sum of the power consumption levels of the compute nodes and the requested increase in power utilization of the first compute node is compared with an allowable power budget for a compute node pool. In addition, the power state of the first compute node power component is varied in response to the comparison.

Claims

1. A method for managing power consumption among compute nodes having respective power components, said power components having a plurality of power states, said method comprising:

determining whether an increase in the power utilization of a first compute node of the compute nodes is detected;

comparing a sum of the power consumption levels of the compute nodes and the requested increase in power utilization of the first compute node with an allowable power budget for a compute node pool containing the compute nodes in response to an increase in the power utilization of the first compute node being determined; and

varying the power state of the first compute node power component in response to the comparison.

2. The method according to claim 1, further comprising:

setting the power states of the compute node power components to minimum power states;

determining whether the first compute node has been assigned to a user;

assigning the power state of the first compute node power component to a level lower than the minimum power state in response to a determination that the first compute node has been assigned to a user; and

performing a computational task on the first compute node.

3. The method according to claim 1, wherein comparing a sum of the power consumption levels of the compute nodes and the increase in power utilization of the first compute node with an allowable power budget for the compute nodes further comprises determining whether the sum of the power consumption levels of the compute nodes and the increase in power utilization of the first compute node falls below the allowable power budget for the compute nodes.

4. The method according to claim 3, wherein varying the power state of the first compute node power component further comprises decreasing the power state of the first compute node power component to thereby increase the power utilization of the first compute node in response to the sum of the power consumption levels of the compute nodes and the increase in power utilization of the first compute node falling below the allowable power budget for the compute node pool.

5. The method according to claim 4, further comprising:

determining whether a decrease in the power utilization of the first compute node power component is detected; and

increasing the power state of the first compute node power component to thereby decrease the power utilization of the first compute node in response to a detected decrease being determined.

6. The method according to claim 3, wherein varying the power state of the first compute node power component further comprises increasing the power state of the first compute node power component to thereby decrease the power utilization of the first compute node in response to the sum of the power consumption levels of the compute nodes and the increase in power utilization of the first compute node exceeding the allowable power budget for the compute node pool.

7. The method according to claim 6, further comprising:

determining whether a current power consumption level of the compute nodes is below the allowable power budget for the compute node pool; and

determining whether a decrease in the power state of the first compute node power component will cause the power consumption level of the compute nodes to exceed the allowable power budget for the compute node pool in response to the current power consumption level falling below the allowable power budget.

8. The method according to claim 7, further comprising:

selecting a next lowest available power state that allows the power consumption level of the compute nodes to remain within the allowable power budget for the compute node pool in response to a determination that a decrease in the power state of the first compute node power component will not cause the power consumption level of the compute nodes to exceed the allowable power budget for the compute node pool; and

decreasing the power state of the first compute node power component to the selected next lowest available power state to thereby increase the power utilization of the first compute node.

9. The method according to claim 6, further comprising:

determining whether a current power consumption level of the compute nodes is below the allowable power budget for the compute node pool;

selecting a compute node from the compute node pool to move to an increased power state in response to the current power consumption level exceeding the allowable power budget;

determining whether the selected compute node comprises the first compute node; and

increasing the power state of the first compute node power component to thereby decrease the power utilization of the first compute node in response to the selected compute node comprising the first compute node.

10. The method according to claim 3, further comprising:

determining whether to decrease the power state of the first compute node in response to the sum of the power consumption levels of the compute nodes and the increase in power utilization of the first compute node falling below the allowable power budget for the compute node pool; and

decreasing the power state of the first compute node in response to a decision to decrease the power state of the first compute node.

11. The method according to claim 1, further comprising:

determining whether a thermal event has occurred on the first compute node; and

increasing the power state of the first compute node power component to a minimum power state in response to a determination that a thermal event has occurred.

12. The method according to claim 11, further comprising:

determining whether the thermal event has cleared; and

performing a computational task on the first compute node in response to a determination that the thermal event has cleared.

13. The method according to claim 1, wherein varying the power state further comprises varying at least one of a voltage and frequency of a processor, varying a power state of a memory disk, varying a spin rate of a memory disk, varying which components of the compute nodes are activated and deactivated, and by using heterogeneity in the components of the compute nodes.

14. The method according to claim 1, wherein varying the power state further comprises varying the power state in a non-incremental manner.

15. A power management agent for managing power consumption of compute nodes having respective power components, said power components having a plurality ofpower states, said power management agent comprising:

a module for determining a power budget limit for the compute nodes;

a module for determining a power consumption level of the compute nodes, said module for determining a power consumption level of the compute nodes being configured to determine at least one or both of a current power consumption level and an increase in a utilization level of a first compute node of the compute nodes;

a module for comparing the power consumption level of the compute nodes and the power budget limit; and

a module for varying the power states of one or more power components associated with the first compute node based upon comparisons made by the module for comparing.

16. The power management agent according to claim 15, wherein the module for comparing is further configured to determine whether a sum of the power consumption levels of the compute nodes and an increase in the utilization level of the first compute node falls below the allowable power budget for the compute nodes.

17. The power management agent according to claim 16, wherein the module for varying is further configured to decrease the power states of the one or more power components to thereby increase the power utilization of the one or more power components in response to the sum of the power consumption levels of the compute nodes and the increase in power utilization level of the first compute node falling below the allowable power budget for the compute nodes.

18. The power management agent according to claim 16, wherein the module for varying is further configured to increase the power states of the one or more power components to thereby decrease the power utilization of the one or more power components in response to the sum of the power consumption levels of the compute nodes and the increase in power utilization level of the first compute node exceeding the allowable power budget for the compute nodes.

19. The power management agent according to claim 18, wherein the module for determining a power consumption level of the compute nodes is further configured to determine whether a current power consumption level of the compute nodes is below the allowable power budget for the compute node pool, wherein the module for varying is further configured to determine whether a decrease in the power state of the first compute node power component will cause the power consumption level of the compute nodes to exceed the allowable power budget for the compute node pool.

20. The power management agent according to claim 19, wherein the module for varying is further configured to select a next lowest available power state that allows the power consumption level of the compute nodes to remain within the allowable power budget for the compute node pool in response to a determination that a decrease in the power state of the first compute node power component will not cause the power consumption level of the compute nodes to exceed the allowable power budget and to decrease the power state of the first compute node power component to the selected next lowest available power state to thereby increase the power utilization of the first compute node.

21. The power management agent according to claim 15, further comprising:

a module for determining whether a thermal event has occurred in one or more of the compute nodes, said module for determining whether a thermal event has occurred being configured to increase the power state of the one or more compute nodes to a minimum power state in response to a determination that a thermal event has occurred.

22. The power management agent according to claim 15, wherein the compute nodes comprise blade PCs.

23. A system for managing power among compute nodes having respective power components, said power components having a plurality of power states, said system comprising:

means for detecting power utilization of the compute nodes, said means for detecting being configured to detect increases in power utilization levels of the compute nodes;

means for comparing the power utilization levels of the compute nodes with an allowable power budget for a compute node pool; and

means for varying the power states of the power components based upon comparisons made by the means for comparing.

24. The system according to claim 23, further comprising:

means for detecting a thermal event in one or more compute nodes; and

wherein the means for varying is further configured to reduce the power states of the one or more compute nodes in response to a detected thermal event in the one or more compute nodes by the means for detecting.

25. A computer program product embodied on a computer-readable medium and comprising code that, when executed, causes a computer to perform the following:

determine whether an increase in power utilization of a first compute node of a plurality of compute nodes is detected;

compare a sum of power consumption levels of the plurality of compute nodes and the requested increase in power utilization of the first compute node with an allowable power budget for a compute node pool containing the compute nodes in response to an increase in the power utilization of the first compute node being determined; and

vary the power state of the first compute node power component in response to the comparison.