Inferred Power Consumption for Computing Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high cost and inaccuracy of independent power monitoring devices for computing devices in data centers, especially when trying to monitor individual devices, make it economically and practically challenging to track power consumption effectively, as installing thousands of devices is expensive and aggregate monitoring loses the ability to distinguish power consumption of specific devices.
Innovation Solution
A method to infer the power consumption of computing devices using utilization-to-power-consumption transfer functions derived from benchmarking individual components, allowing for aggregate monitoring that can distinguish power usage of individual devices and detect potential failures or abnormalities by comparing current and historical data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If independent power monitoring devices are installed for each computing device, then power consumption measurement precision is improved, but device cost and system complexity increase significantly
Solution Approach 1:
The patent creates a virtual copy of the power monitoring functionality by using software agents on computing devices to report power consumption data to a centralized server. Instead of installing physical monitoring hardware on each device, the system uses software-based power consumption models that replicate the measurement capability, thereby maintaining measurement precision while avoiding the complexity and cost of thousands of physical monitoring devices.
2Measurement precision
If independent power monitoring devices are installed for each computing device, then individual device power consumption monitoring is improved, but installation cost increases
Solution Approach 1:
The patent replaces expensive physical power monitoring devices with software-based power consumption models. Each computing device runs an agent that collects power consumption data and transmits it to a centralized server, creating a virtual monitoring system that maintains individual device monitoring capability while dramatically reducing installation and hardware costs.
Solution Approach 2:
The patent substitutes mechanical/physical power monitoring hardware with a software-based system. Instead of using physical sensors and monitoring devices to measure power consumption, the system uses software agents that collect data from device metrics and power consumption models, thereby eliminating the need for expensive physical infrastructure while maintaining monitoring accuracy.
3Ease of manufacture
If aggregate power monitoring is used instead of individual monitoring, then device cost is reduced, but the ability to distinguish individual device power consumption is lost
Solution Approach 1:
The patent segments the power consumption monitoring system into two parts: a centralized server that aggregates data and individual device agents that collect and transmit their specific power consumption data. This segmentation allows the system to maintain low cost through centralized processing while preserving individual device information through distributed data collection, resolving the contradiction between cost reduction and information retention.
Data Source
AI summary
The power consumption of a computing device is inferred from the utilization rates of individual components of the computing device and a utilization-to-power-consumption transfer function that was derived by benchmarking that, or an analogous, computing device. The inferred power consumption of a computing device is aggregated to infer the power consumption of various groups and super-groups of computing devices. The historical power consumption of computing devices is inferred based on the utilization rates of individual components of the computing devices at relevant times in the past. Historical power consumption is used to derive a power consumption profile of a computing device and the inferred current power consumption of such a computing device is compared to such a power consumption profile, and to the historical power consumption, to identify deviations therefrom, which can provide proactive detection of potential hardware faults, software glitches, or other errors.