Non-Intrusive Server Power Management via Utilization Prediction

Solution Overview

Problem

Data centers face high power consumption due to idle servers, which contribute to significant operational expenses, as they draw substantial power even when not fully loaded, and existing power management solutions are intrusive and require additional software or hardware installations.

Innovation Solution

A non-intrusive power management system that uses statistical analysis to predict server utilization based on load changes, allowing centralized decision-making to turn servers on or off without custom software or hardware installation, utilizing existing OS information and standardized protocols like SNMP or WMI, and employing a staggered suspend/resume approach to manage server states between S0 and S5.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If servers are kept running to maintain high availability, then service reliability is improved, but power consumption increases

Engineering Contradiction:

Improveservice availabilityVSAvoidpower consumption

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system performs preliminary characterization of server utilization patterns by collecting and analyzing utilization information before making power management decisions. This allows the system to predict future utilization and make proactive power state changes that maintain availability while reducing energy consumption during low-demand periods

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system continuously collects utilization information from servers, analyzes it to identify patterns, and uses this feedback to adjust server power states. The characterization process creates a closed-loop system where past utilization data informs future power management decisions, enabling dynamic adaptation to changing load conditions

Inventive Principle:

Principle #23Feedback

2Use of energy by moving object

If servers are shut down to reduce power consumption, then energy costs are reduced, but service availability deteriorates

Engineering Contradiction:

Improvepower consumptionVSAvoidservice availability

Core Design Contradiction:

Use of energy by moving objectVSReliability

Solution Approach 1:

The system performs preliminary characterization of server utilization patterns by collecting and analyzing utilization information before making power management decisions. This allows the system to predict future utilization and make proactive power state changes that maintain availability while reducing energy consumption during low-demand periods

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system continuously collects utilization information from servers, analyzes it to identify patterns, and uses this feedback to adjust server power states. The characterization process creates a closed-loop system where past utilization data informs future power management decisions, enabling dynamic adaptation to changing load conditions

Inventive Principle:

Principle #23Feedback

3Use of energy by moving object

If intrusive power management solutions are implemented, then power consumption is reduced, but system complexity increases

Engineering Contradiction:

Improvepower consumptionVSAvoidsoftware installation requirements

Core Design Contradiction:

Use of energy by moving objectVSDevice complexity

Solution Approach 1:

The system uses self-service mechanisms by leveraging existing operating system information and standardized protocols that are already present on servers. The power management function collects utilization data through standard OS interfaces and network protocols without requiring custom software installations, thereby reducing complexity while achieving power savings

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system employs universal standardized protocols (SNMP, WMI) that serve multiple purposes across different server platforms. These protocols are already integrated into the server infrastructure for monitoring and management, so reusing them for power management decisions avoids adding new complex software layers while achieving cross-platform power optimization

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS8850243B2Non-intrusive power management

Publication Date: 2014.09.30 AMAZON TECH INC

AI summary

A method and system for managing power consumption of a pool of computing devices that are logically grouped to provide a common set of functionality is disclosed. One aspect of certain embodiments includes predicting resource utilization for each device without installing customized software, firmware or hardware on the device.