Method for managing power consumption of multiple computer servers

• US 20030177165A1
• Filed: 03/18/2002
• Published: 09/18/2003
• Est. Priority Date: 03/18/2002
• Status: Active Grant

• Current Assignee: Lenovo International Limited (Lenovo Group Ltd.)
• Original Assignee: International Business Machines Corporation
• Inventors: Harper, Richard Edwin, Hunter, Steven Wade, Bradley, David John
• Granted Patent: US 6,795,928 B2
• Time in Patent Office: Days
• US Class Current: 718/105
• CPC Class Codes: G06F 1/206 comprising thermal management
• CPC Class Codes: G06F 1/3206 Monitoring of events, devices or parameters that trigger a change in power modality
• CPC Class Codes: H05K 7/20836 Thermal management, e.g. server temperature control
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A method for managing power consumption of multiple computer servers is disclosed. The current workload information and the current capacity information of a group of computer servers within a computer server pool are determined. A short term history and a long term history of the workload information are then built based on the determined workload information. Next, the information regarding short-term variations in workload is determined. Similarly, the information regarding long-term variations in workload is determined. Based on a combination of the above-mentioned determined short-term and long-term information, at least one of the computer servers is powered on or powered off.

Claims

1. A method for managing power consumption in computer servers, said method comprising:

determining current workload information and current capacity information on a plurality of computer servers;

building a short-term history and a long-term history of workload information based on said determined workload information;

determining information regarding short-term variations in workload;

determining information regarding long-term variations in workload; and

powering on or powering off at least one of said plurality of computer servers according to a combination of said determined short-term and long-term information.

2. The method of claim 1, wherein said short-term variations in workload is determined by a gain-based algorithm.

3. The method of claim 1, wherein said long-term variations in workload is determined by an algorithm based on temporal characterization.

4. A computer program product residing on a computer usable medium for managing power consumption in computer servers, said computer program product comprising:

program code means for determining current workload information and current capacity information on a plurality of computer servers;

program code means for building a short-term history and a long-term history of workload information based on said determined workload information;

program code means for determining information regarding short-term variations in workload;

program code means for determining information regarding long-term variations in workload; and

program code means for powering on or powering off at least one of said plurality of computer servers according to a combination of said determined short-term and long-term information.

5. The computer program product of claim 4, wherein said short-term variations in workload is determined by a gain-based algorithm.

6. The computer program product of claim 4, wherein said long-term variations in workload is determined by an algorithm based on temporal characterization.

7. A method for managing power consumption in computer servers, said method comprising:

selecting a sample window having a plurality of workload sample;

determining a maximum sample-to-sample deviation value of said plurality of workload samples within said selected sample window;

determining a capacity envelope; and

adjusting a current capacity of said plurality of computer servers if said current capacity falls outside said capacity envelope.