System and method for managing energy consumption in a compute environment

• US 8,549,333 B2
• Filed: 09/18/2012
• Issued: 10/01/2013
• Est. Priority Date: 04/21/2008
• Status: Active Grant

• Current Assignee: Seagate Technology LLC (Seagate Technology Holdings Plc)
• Original Assignee: Adaptive Computing Enterprises, Inc. (Ala Services LLC)
• Inventors: Jackson, David Brian
• Primary Examiner(s): CONNOLLY, MARK A
• Application Number:
  US13/621,989
• Publication Number:
  US 20130024710A1
• Time in Patent Office: 378 Days
• Field of Search: 713/300, 713/320, 713/322, 713/324, 718/104, 718/105
• US Class Current: 713/300
• CPC Class Codes: G05D 23/1917 using digital means
• CPC Class Codes: G06F 1/3203 Power management, i.e. event-based initiation of a power-saving mode
• CPC Class Codes: G06F 1/3206 Monitoring of events, devices or parameters that trigger a change in power modality
• CPC Class Codes: G06F 1/3256 Power saving in optical drive
• CPC Class Codes: G06F 1/3268 Power saving in hard disk drive
• 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: G06F 1/329 by task scheduling
• CPC Class Codes: G06F 9/5088 involving task migration
• CPC Class Codes: G06F 9/5094 where the allocation takes into account power or heat criteria power management in computers in general G06F1/3203; thermal management in computers in general G06F1/206
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A system and method for reducing power consumption in clusters, grids, on-demand centers, etc. These principles can reduce both direct and indirect power consumption while maintaining either full cluster performance or adequate SLA based cluster performance. The method includes receiving at least one state data point regarding power consumption or temperature of at least one resource within the compute environment. Using intelligent policies to control power consumption, the method implements and interfaces with power managements facilities within the cluster, grid or on-demand center to implement policies, make dynamic changes, make predictions or actions, etc., to reduce the direct or indirect power consumption associated with a compute environment. Methods can include analysis of current workload and/or future workload in taking energy saving actions, and also involve reporting state information and updating algorithms based on historical experience or outside sources of information.

Claims

1. A method comprising:

receiving data about a current state of a compute environment, wherein the compute environment comprises a plurality of nodes under common management in which a workload manager schedules and distributes workload among the plurality of nodes and wherein each node of the plurality of nodes is an independent computer that is networked with other nodes of the plurality of nodes in the compute environment;

analyzing the workload to yield an analysis; and

migrating, based at least in part on the data and the analysis and in a manner related to energy consumption, a reservation from a first time associated with a first group of resources to a second time, to yield a second reservation of a second group of resources.

2. The method of claim 1, wherein the data is at least one of a temperature and power consumption.

3. The method of claim 1, further comprising:

analyzing future workload that will be consumed in the compute environment, to yield a second analysis wherein migrating the reservation is based on the data and the analysis of the current workload and the second analysis.

4. The method of claim 1, further comprising modifying a use of at least one resource in the compute environment by powering on or off the at least one resource.

5. A method of managing power consumption, the method comprising:

receiving a current state of a compute environment, wherein the compute environment comprises a plurality of nodes under common management in which a workload manager schedules and distributes workload among the plurality of nodes and wherein each node of the plurality of nodes is an independent computer that is networked with other nodes of the plurality of nodes in the compute environment;

analyzing at least a portion of the workload to yield an analysis;

predicting at least one power consumption saving action based on the current state and the analysis; and

implementing, based on the predicting, a predicted at least one power consumption saving action in the compute environment, wherein the power consumption saving action comprises migrating a first reservation of first compute resources in the compute environment at a first time to yield a second reservation of second compute resources in the compute environment at a second time.

6. The method of claim 5, wherein the at least one power consumption saving action is at least one of:

powering down a node, powering down memory, spinning down a disk, lowering a clock speed of a processor, powering down a hard drive, and placing a resource in a low power consumption mode.

7. The method of claim 5, further comprising:

analyzing the compute environment and workload as the workload consumes resources in the compute environment to yield a second analysis; and

adjusting the predicted at least one power consumption saving action based on the second analysis.

8. The method of claim 7, wherein the adjusting further comprises one of:

(1) increasing or decreasing a number of powered down nodes from an implemented amount;

or

(2) increasing or decreasing an amount of powered down memory from an implemented amount.

9. A method of managing power consumption, the method comprising:

receiving a current power consumption state of a compute environment, wherein the compute environment comprises a plurality of nodes under common management in which a workload manager schedules and distributes workload among the plurality of nodes and wherein each node of the plurality of nodes is an independent computer that is networked with other nodes of the plurality of nodes in the compute environment;

analyzing queued jobs scheduled to consume resources in the compute environment to yield an analysis;

predicting, based on the analysis, power consumption when at least one of the queued jobs is consumed in the compute environment; and

consuming the at least one job in the compute environment with at least one power consumption saving action implemented based on the predicting, wherein the power consumption saving action comprises migrating a first reservation of first compute resources in the compute environment at a first time to yield a second reservation of second compute resources in the compute environment at a second time.

10. The method of claim 9, wherein the at least one power consumption saving action is one of:

job migration within the compute environment, job migration to a second compute environment, adjusting a cooling system, and adjusting power consumption of at least one resource in the compute environment.

11. The method of claim 9, wherein the at least one power consumption saving action relates to modifying use of a cooling facility associated with the compute environment.

12. The method of claim 11, wherein modifying the cooling facility includes pre-cooling resources in the compute environment prior to consuming the at least one job.

13. The method of claim 9, wherein the at least one power consumption saving action is modifying a data prestaging reservation.