Power Consumption Measurement for Virtualized Applications
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional charge-back mechanisms in cloud computing do not accurately capture power utilization, leading to inefficiencies in cost allocation and increased costs for data center operations, as they fail to account for the power consumption and carbon emissions associated with virtualized applications.
Innovation Solution
A method to determine the share of power consumption by an application executing on a server by obtaining metrics related to server operation and resource utilization, allowing for the calculation of power costs and carbon emissions, and enabling the migration of virtual machines to optimize energy usage and costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional charge-back mechanisms are used in cloud computing, then cost allocation is simplified, but power utilization measurement is inaccurate
Solution Approach 1:
The patent segments power consumption measurement into multiple components: total power consumption of the physical server, power consumption attributed to virtual machines, and power consumption attributed to specific applications within those virtual machines. This segmentation enables accurate measurement of power utilization at different levels of the virtualized infrastructure, resolving the contradiction between measurement precision and mechanism complexity by breaking down the measurement task into manageable segments.
2Measurement precision
If power consumption measurement is implemented at the application level in virtualized environments, then cost allocation accuracy is improved, but system complexity increases
Solution Approach 1:
The patent introduces intermediary components including power management modules, virtual machine monitoring agents, and application performance monitoring tools that act as mediators between the physical infrastructure and the charge-back system. These intermediaries collect, process, and translate raw data into meaningful power consumption metrics, improving measurement accuracy while managing system complexity through standardized interfaces and modular architecture.
Solution Approach 2:
The patent replaces traditional mechanical or manual power measurement approaches with electronic and software-based monitoring systems. This includes using electronic sensors to measure power consumption, software agents to collect resource utilization metrics, and automated algorithms to calculate and allocate power costs, thereby improving measurement precision while reducing the complexity associated with manual processes.
3Productivity
If comprehensive power metrics are collected across CPU, memory, network, and cooling, then energy management is optimized, but data collection complexity increases
Solution Approach 1:
The patent implements a universal power management framework that simultaneously collects and processes metrics across multiple domains including CPU power consumption, memory power consumption, network power consumption, and cooling power consumption. This multi-functional system uses a common data collection architecture and standardized processing methods that improve energy management efficiency while managing complexity through unified approaches rather than separate systems for each metric type.
Data Source
AI summary
Example embodiments of the present invention provide a method, apparatus and computer program product for determining a share of power consumption by an application executing on a server. The method includes obtaining metrics relating to operation of the server and obtaining metrics relating to server resource utilization attributable to the application. A transformation is then performed using the metrics relating to operation of the server and server resource utilization attributable to the application to determine the respective share of power consumption by the application executing on the server. Additionally, greenhouse-gases attributable to the application also may be calculated.