Managing power consumption and performance of computing systems

Patent metadata

Field	Value
Publication number	US-9857858-B2
Application number	US-201213976817-A
Country	US
Kind code	B2
Filing date	May 17, 2012
Priority date	May 17, 2012
Publication date	Jan 2, 2018
Grant date	Jan 2, 2018

Abstract

A method and system for managing power consumption and performance of computing systems are described herein. The method includes monitoring an overall power consumption of the computing systems to determine whether the overall power consumption is above or below an overall power consumption limit, and monitoring a performance of each computing system to determine whether the performance is within a performance tolerance. The method further includes adjusting the power consumption limits for the computing systems or the performances of the computing systems such that the overall power consumption is below the overall power consumption limit and the performance of each computing system is within the performance tolerance.

First claim

What is claimed is:

1. A method for managing power consumption and performance of computing systems, comprising:
monitoring an overall power consumption of the computing systems to determine whether the overall power consumption is above or below an overall power consumption limit for a particular workload, wherein the overall power consumption is based on, at least in part, and socket level power consumption and each computing system comprises a node management agent chassis that is to determine a corresponding socket level power consumption limit;
monitoring a performance of each computing system to determine whether a performance of each computing system is within a performance tolerance, wherein the performance is a response of the computing systems during and after execution of the particular workload, and wherein the response is a number of transactions completed; and
adjusting power consumption limits for the computing systems or performances of the computing systems such that the overall power consumption of the computing systems is below the overall power consumption limit and the performance of each computing system is within the performance tolerance.

2. The method of claim 1 , wherein the method is executed via a data center management agent within a remote computing system.

3. The method of claim 2 , comprising sending the power consumption limit for a computing system from the data center management agent within the remote computing system to a node management agent within the computing system, the node management agent being configured to maintain a power consumption of the computing system below the power consumption limit and a performance of the computing system within the performance tolerance.

4. The method of claim 1 , wherein the computing systems comprise homogeneous computing systems with an identical hardware configuration and an identical software configuration.

5. The method of claim 4 , comprising uniformly adjusting performances of the computing systems such that the computing systems are operating at an identical performance and the overall power consumption is below the overall power consumption limit.

6. The method of claim 4 , comprising:
periodically monitoring the overall power consumption of the computing systems;
in response to determining that the overall power consumption exceeds an upper threshold power consumption, decreasing the performances of the computing systems such that the overall power consumption is below the upper threshold power consumption; and
in response to determining that the overall power consumption is below a lower threshold power consumption, increasing the performances of the computing systems such that the overall power consumption is above the lower threshold power consumption.

7. The method of claim 4 , comprising:
determining a performance and a maximum power consumption for each performance state of a device within the computing systems from an Advanced Configuration and Power Interface (ACPI) performance state table; and
determining a target performance state for the computing systems based on the performance and the maximum power consumption for each performance state, wherein the target performance state is determined such that the overall power consumption is below the overall power consumption limit and the performances of the computing systems are within the performance tolerance.

8. The method of claim 4 , comprising:
determining a power consumption limit for each computing system by dividing the overall power consumption limit by a total number of computing systems;
monitoring the computing systems to determine a power consumption and a performance of each computing system over a specified period of time; and
adjusting the power consumption limit for each computing system based on the power consumption and the performance over the specific period of time, wherein the power consumption limit for each computing device is adjusted such that performances of the computing systems are maintained within the performance tolerance.

9. The method of claim 4 , comprising:
operating the computing systems at a uniform power consumption;
monitoring the computing systems to determine a performance of each computing system over a specified period of time;
determining a specified number of computing systems with highest performances;
powering down each computing system that is not within the specified number of computing systems with the highest performances; and
adjusting the power consumption limit for each of the specified number of computing systems based on the performance of each computing system over the specific period of time.

10. The method of claim 4 , comprising:
operating the computing systems at a uniform performance;
monitoring the computing systems to determine a power consumption of each computing system over a specified period of time;
determining a specified number of computing systems with lowest power consumptions;
powering down each computing system that is not within the specified number of computing systems with the lowest power consumptions; and
adjusting a performance of each of the specified number of computing systems based on the power consumption of each computing system over the specific period of time.

11. The method of claim 1 , comprising:
operating each computing system in each of a plurality of performance states, wherein each performance state comprises a performance and a maximum power consumption;
collecting data relating to a power consumption and a performance of each computing system in each performance state;
building a table comprising measured performance states based on the data, wherein each of the measured performance states comprises an average performance and an average power consumption; and
using the table to determine the power consumption limit for each computing system, wherein the power consumption limit is determined such that the performance of the computing system is within the performance tolerance.

12. The method of claim 1 , comprising:
monitoring a power consumption of each computing system and a time when each computing system reaches a specific boundary condition;
increasing power consumption limits for a subset of the computing systems that reach the specific boundary condition in a slowest time period; and
decreasing power consumption limits for a remaining number of the computing systems that reach the specific boundary condition in a fastest time period.

13. A system for managing power consumption and performance of computing systems, comprising:
a data center management system, wherein the data management system is a computing device configured to determine a platform level power consumption limit for each of a plurality of computing systems in a data center for a particular workload based on an overall power consumption limit and a performance tolerance for the plurality of computing systems, wherein the performance tolerance is based, in part, on a response of the plurality of computing systems during and after execution of the particular workload; and
the plurality of computing systems, wherein each computing system comprises:
a node management agent, wherein the node management agent is a chassis comprising multiple blades that is configured to maintain a platform level power consumption of the computing system below the platform level power consumption limit by determining a socket level power consumption limit for each socket within a platform of the computing system; and
a plurality of component management agents, wherein each component management agent is configured to maintain the soc

Assignees

• Bodas Devadatta V
• Crawford John H
• Gara Alan G
• Intel Corp

Inventors

• Bodas Devadatta V
• Crawford John H
• Gara Alan G

Classifications

• G06F1/3234

  Power saving characterised by the action undertaken · CPC title

• G06F1/3206 Primary

  Monitoring of events, devices or parameters that trigger a change in power modality · CPC title

• G06F1/3209

  Monitoring remote activity, e.g. over telephone lines or network connections · CPC title