Control of a computing system having adjustable inputs

• US 9,143,554 B2
• Filed: 10/13/2008
• Issued: 09/22/2015
• Est. Priority Date: 10/13/2008
• 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: Liu, Xue, Zhu, Xiaoyun, Wang, Zhikui, Singhal, Sharad, Cohen, Ira
• Primary Examiner(s): Pezzlo, John
• Application Number:
  US12/250,256
• Publication Number:
  US 20100091786A1
• Time in Patent Office: 2,535 Days
• Field of Search: 370/252, 370/386, 370/419
• US Class Current: 1/1
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46

Abstract

A technique for controlling an output of a computing system having multiple adjustable inputs includes providing a set of adjustable inputs to the computing system, observing an output of the computing system while the system is in operation, and selecting a subset of adjustable inputs from the set of adjustable inputs based on the observation of the output. The inputs in the selected subset are then adjusted to achieve a desired output of the computing system.

Claims

1. A method of controlling an output of a computing system, comprising:

providing, via a processing system, a set of adjustable inputs to the computing system;

observing, via the processing system, while in operation, performance metrics associated with an output of the computing system;

identifying, via the processing system, adjustable inputs from the set of adjustable inputs which most affect at least a selected one of the performance metrics, the identifying comprising applying a statistical analysis to the observed performance metrics to identify the adjustable inputs; and

adjusting, via the processing system, while in operation, only the identified adjustable inputs to achieve a desired value of the at least selected one perfbrrnance metric without adjusting the adjustable inputs that are not identified but are provided to the computing system.

2. The method as recited in claim 1, wherein identifying the adjustable inputs is further based on observing, via the processing system, a response of the output of the computing system to adjustment of the provided adjustable inputs.

3. The method as recited in claim 1, further comprising:

generating via the processing system, a reduced input-output model of the computing system using the identified adjustable inputs; and

adjusting the identified adjustable inputs based on the reduced input-output model.

4. The method as recited in claim 3, further comprising:

periodically updating the reduced input-output model via the processing system.

5. The method as recited in claim 3, further comprising:

identifying, via the processing system, a new subset of adjustable inputs in response to a change of the output; and

re-generating, via the processing system, the reduced input-output model using the new subset of adjustable inputs.

6. The method of claim 1, further comprising providing information technology services via the processing system, wherein the observing task, the identifying task, and the adjusting task are part of the information technology services.

7. A system for dynamically controlling an output of a computing system having a plurality of adjustable inputs, comprising:

an input selection module to dynamically identify, via a processor, a subset of adjustable inputs from a plurality of adjustable inputs that are provided to a computing system in response to observation of performance metrics associated with an output of the computing system while the computing system is in operation, wherein the input selection module applies a statistical analysis to identify which adjustable inputs of the set of adjustable inputs most affect at least a selected one of the performance metrics while the computing system is in operation; and

a control module coupled to the input selection module and configured to adjust, via the processor, only the adjustable parameters in the identified subset to achieve a desired value of the at least one performance metric.

8. The system as recited in claim 7, wherein the control module comprises a model identification module to generate, via the processor, an input-output model of the computing system based on the identified subset.

9. The system as recited in claim 8, wherein the model identification module periodically regenerates the input-output model via the processor.

10. The system as recited in claim 7, wherein the input selection module implements, via the processor, a statistical regression analysis to determine which adjustable inputs have the most affect on the at least one performance metric.

11. The system as recited in claim 7, wherein the computing system is a shared server system.

12. The system as recited claim 11, wherein the plurality of adjustable inputs includes at least one of an application-level parameter and a system-level parameter.

13. The system as recited in claim 7, wherein the input selection module identifies, via the processor, the adjustable inputs that most affect the at least selected one of the performance metrics in response to an observed change in the output of the computing system to adjustment of the plurality of provided adjustable inputs.

14. The system as recited in claim 13, wherein the model identification module regenerates, via the processor, the input-output model based on the new subset of the adjustable inputs.

15. An article comprising at least one non-transitory computer-readable storage medium containing instructions that when executed by a processor-based system cause the processor-based system to:

observe performance metrics associated with an output of a computing system during operation of the computing system, the computing system having a plurality of adjustable inputs;

identify a subset of adjustable inputs of the plurality of adjustable inputs which most affect at least one of the performance metrics, the identification comprising applying a statistical analysis, during operation of the computing system, to identify the subset of adjustable inputs;

generate an input-output model of the computing system based on the determined subset of adjustable inputs; and

use the input-output model to adjust, during operation of the computing system, only the adjustable inputs in the determined subset to achieve a desired output of the computing system without adjusting the adjustable inputs of the computing system that are not in the determined subset.

16. The article of claim 15, the storage medium storing instructions that when executed by the processor-based system cause the processor-based system to identify the subset of adjustable inputs based on an observation of a response of the output of the computing system to adjustment of the plurality of adjustable inputs.