Rack Airflow Deficiency Detection and Automated Cooling Remediation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solutions fail to effectively identify and remediate airflow deficiencies in server racks, particularly due to physical obstructions or air voids caused by neighboring devices, leading to inadequate cooling and potential overheating, which is difficult to manage even with manual inspections.
Innovation Solution
A software-defined infrastructure, including a rack device manager that discovers rack devices, calculates airflow metrics, and generates classifiers to detect airflow deficiencies, allowing for automatic remediation by adjusting fan speeds, workload distribution, or applying power caps to neighboring devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual inspection is used to identify airflow deficiencies, then physical obstructions can be identified, but air voids are difficult to identify and cooling management remains inadequate
Solution Approach 1:
The patent replaces manual mechanical inspection with an automated software-defined infrastructure that uses airflow metrics, machine learning classifiers, and software-controlled remediation actions to detect and resolve airflow deficiencies, eliminating the need for manual rack inspections
Solution Approach 2:
The patent introduces a rack device manager as an intermediary system that collects airflow metrics from rack devices, processes them through machine learning classifiers to identify airflow deficiency scenarios, and automatically executes remediation actions, serving as a mediator between monitoring and cooling management
2Temperature
If fan speed is increased to cool a rack device, then cooling capacity improves, but energy consumption increases and the root cause of inadequate cooling is not addressed
Solution Approach 1:
The patent performs preliminary identification of airflow deficiency scenarios using machine learning classifiers before executing remediation actions, allowing the system to address the root cause of inadequate cooling (such as air voids or obstructions) before resorting to increasing fan speed, thereby avoiding unnecessary energy consumption
Solution Approach 2:
The patent enables the cooling system to self-diagnose airflow deficiencies using machine learning models and self-correct by automatically executing remediation actions (such as adjusting rack device positions or removing obstructions), eliminating the need for manual intervention and optimizing energy usage
Data Source
AI summary
A software-defined infrastructure can identify and remediate an airflow deficiency scenario on a rack device. A rack device manager can be configured to discover rack devices and create a representation of their physical locations. The rack device manager can also be configured to periodically retrieve airflow metrics of the rack devices to calculate an estimated airflow for each rack device. The rack device manager can use the estimated airflows and the airflow metrics to generate a rack device classifier for each rack device. Using these rack device classifiers, the rack device manager can detect when rack devices are experiencing airflow deficiencies and attempt to automatically remediate such deficiencies.