Environmentally cognizant power management

• US 8,671,294 B2
• Filed: 03/07/2008
• Issued: 03/11/2014
• Est. Priority Date: 03/07/2008
• Status: Active Grant

• Current Assignee: Sunbird Software, Inc.
• Original Assignee: Raritan Americas Incorporated (Legrand SA)
• Inventors: Malik, Naim R., Onyshkevych, Vsevolod, Paetz, Christian, Somasundaram, Siva, Weinstock, Neil, Yang, Allen
• Primary Examiner(s): Dang, Khanh
• Application Number:
  US12/044,530
• Publication Number:
  US 20090228726A1
• Time in Patent Office: 2,195 Days
• Field of Search: 713/300, 713/320, 713/321, 713/323, 713/340, 700/295
• 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 1/329 by task scheduling
• CPC Class Codes: G06F 9/505 considering the load
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A method for environmentally cognizant power management that gathers server, application and environmental information from different devices to compute an aggregate behavior model.

Claims

1. A method for environmentally cognizant power management in a computing environment, comprising the steps of:

gathering first values for a first level of server power metrics from one or more of servers and power supply units in the computing environment;

gathering first values for a second level of server power metrics from operating system and applications executed within the servers;

gathering first values for environmental metrics;

generating a predictive behavior power utilization model based on at least the first level of server power metrics, the second level of server power metrics and the environmental metrics;

distributing an application load in the computing environment based on applying the predictive behavior power utilization model to application, utilization and environmental contexts;

predicting a resulting power utilization according to the predictive behavior power utilization model and the distributed load;

determining an actual power utilization by gathering second values for the first level of server power metrics and the second level of server power metrics under the application load; and

updating the predictive behavior power utilization model according to actual power utilization when a difference between the actual power utilization and the predicted power utilization exceeds a predetermined threshold.

2. The method according to claim 1, wherein the application load is distributed to both physical locations and logical locations.

3. The method according to claim 1, further comprising the step of constant monitoring and analyzing the computing environment for power management decisions.

4. The method according to claim 1, wherein the first level of server power metrics, and the second level of server power metrics are gathered from different mechanisms to improve the accuracy of the power metric computation.

5. The method according to claim 1, wherein the behavior model is dynamic.

6. The method according to claim 1, further including the step of temporally synchronizing the first level of server metrics and the second level of server metrics.

7. The method according to claim 1, further including the step of temporally synchronizing the first level of server metrics, the second level of server metrics and the third level of server metrics.

8. The method according to claim 1, wherein a first sampling interval is used to gather data for long term trending analysis and a second sampling interval is used to gather data for localized short term bursty activity.

9. The method according to claim 1, further comprising the step of:

reallocating the distributed application load according to the adjusted predictive behavior model for improved power utilization.

10. The method according to claim 1, further comprising the steps of:

adjusting one or more of a heating function or a cooling function; and

adjusting a power utilization component of the predictive behavior power utilization model that predicts a power utilization of the adjusted heating function or cooling function.

11. A method for environmentally cognizant power management in a distributed computing system, comprising the steps of:

gathering first values for process level information from a plurality of sources in the distributed computing system;

gathering first values for environmental metrics for the distributed computing system;

generating a predictive behavior power utilization model based on the process level information and the environmental metrics;

allocating an application load among computing elements in the distributed computing system based on applying the behavior model to application, utilization and environmental contexts;

predicting a resulting power utilization according to the predictive behavior power utilization model and the application load;

determining actual power utilization information for the distributed computing system by gathering second values for the process level information and the environmental metrics; and

updating the predictive behavior power utilization model according to the actual power utilization information when a difference between the actual power utilization and the predicted power utilization exceeds a predetermined threshold.

12. The method according to claim 11, wherein the application load is distributed to both physical locations and logical locations.

13. The method according to claim 11, further comprising the step of constant monitoring and analyzing the computing environment for power management decisions.

14. The method according to claim 11, wherein the step of generating uses prioritized process level information.

15. The method according to claim 14, further comprising the step of filtering the process level information to prioritize the process level information with respect to power utilization.

16. Apparatus for environmentally cognizant power management in a computing system, comprising:

a processor that is configured;

to generate a predictive behavior power utilization model based on first values of process level information and environmental metrics received from one or more intelligent power management devices,

to allocate an application load among computing elements in the computing system based on applying the predictive behavior model to application, utilization and environmental contexts,

to predict a resulting power utilization according to the predictive behavior power utilization model and the application load;

to determine actual power utilization information by gathering second values for the process level information and the environmental metrics from the computing system, and

to update the predictive behavior power utilization model according to the actual power utilization information when a difference between the actual power utilization and the predicted power utilization exceeds a predetermined threshold.

17. The apparatus for claim 16, wherein the processor further comprises a filter module that prioritizes the process level information with respect to power utilization.

18. A framework for environmentally cognizant energy conservation, comprising:

an intelligent configuration module that controls information flow and capture pertaining to electrical and computing energy from respective electrical and computing power entities;

a monitor engine in communication with the intelligent configuration module for collecting multiple levels of information pertaining to the electrical and computing energy as directed by the intelligent configuration module; and

an analysis engine coupled to the monitor engine that adaptively tracks an association between electrical and computing power entities based on the information pertaining to electrical and computing energy and environmental information pertaining to the electrical and computer power entities, and adaptively updates a predictive behavior power utilization model based on the information pertaining to the electrical and computing energy when a difference between an actual power utilization and a predicted power utilization according to the predictive behavior power model exceeds a predetermined threshold.

19. The framework of claim 18, wherein the intelligent configuration module and monitor engine collects measurements at predetermined intervals according to defined sampling rates.