Data Center Power Efficiency Modeling via Sensor-Based Energy Estimation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

As data centers perform increasing computations and processes, their power usage and costs rise, necessitating an efficient means to estimate energy usage and enhance efficiency.

Innovation Solution

An apparatus and method that utilize power sensors to measure and model energy consumption, determining data-to-Watt ratios and estimated execution times to optimize power efficiency through domain partitioning, load parallelization, and workload scheduling.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If computation and processes in data centers are increased, then productivity and computational capacity are improved, but power usage and energy consumption increase

Engineering Contradiction:
Improvecomputational capacityVSAvoidpower usage
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent applies parameter changes by dynamically adjusting operational parameters of processing elements based on workload characteristics. The system monitors power consumption and computational output to identify optimal parameter settings that maximize productivity per unit of energy consumed, thereby resolving the contradiction between increased computational capacity and increased power usage.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If power measurements and energy modeling are implemented, then energy consumption estimation precision is improved, but device complexity and measurement infrastructure requirements increase

Engineering Contradiction:
Improveenergy consumption estimationVSAvoidmeasurement infrastructure
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary modeling layer that translates complex power measurements into simplified energy consumption estimates. Rather than directly measuring and analyzing all raw power data from multiple sensors, the system uses computational models as intermediaries to process measurements from select power sensors, thereby achieving precise energy estimation without proportionally increasing infrastructure complexity.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Measurement precision

If power sensors and measurement systems are added, then energy monitoring capability is improved, but device complexity and cost increase

Engineering Contradiction:
Improvepower measurement capabilityVSAvoidsensor infrastructure
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies universality by designing power sensors and measurement systems that serve multiple functions simultaneously. The same measurement infrastructure is used not only for energy consumption monitoring but also for performance characterization, thermal analysis, and computational efficiency evaluation, thereby reducing the need for dedicated sensors for each measurement type and lowering overall system complexity.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS10401933B2Approach for end-to-end power efficiency modeling for data centers
Publication Date: 2019.09.03 FUTUREWEI TECHNOLOGIES INC
  • US10401933B2 patent drawing
  • US10401933B2 patent drawing
  • US10401933B2 patent drawing

AI summary

An apparatus comprising a receiver configured to receive a plurality of power measurements from a plurality of power sensors, and a processor coupled to the receiver and configured to determine an amount of power used by a processing element in a data center by determining a summation of the plurality of power measurements, determine a data to Watt ratio that indicates an amount of data that may be processed by the processing element per unit of power used, determine an estimated execution time for processing the amount of data by the processing element, and determine an estimated energy consumption that indicates an amount of energy to be used by the processing element to process the amount of data.