Liquid Cooling Abnormality Detection for Localized Server Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In information processing systems, particularly supercomputer systems, liquid cooling system abnormalities can lead to system shutdowns, making it difficult to maintain service reliability and hardware monitoring is inadequate, especially in high-demand applications like artificial intelligence and big data analysis.
Innovation Solution
An information processing apparatus with a controller that acquires sensor data and abnormality information from multiple sensors to detect and analyze liquid cooling system issues, determining the range of influence of abnormalities and controlling the system accordingly, allowing for targeted action rather than widespread shutdowns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If liquid cooling system abnormalities are detected, then system reliability is improved through targeted control, but device complexity increases due to multiple sensors and controllers
Solution Approach 1:
The system divides the information processing system into multiple independent information processing apparatuses, each with its own controller and sensors. This segmentation allows localized detection and control of liquid cooling abnormalities without affecting the entire system, improving reliability while managing complexity through modular architecture.
Solution Approach 2:
Controllers in each information processing apparatus acquire sensor data and detect abnormalities in advance before they affect other apparatuses. By performing preliminary detection and control actions locally, the system prevents widespread shutdowns and maintains overall reliability without requiring complex centralized monitoring.
2Reliability
If widespread shutdowns are implemented upon detecting liquid cooling abnormalities, then reliability is maintained, but productivity decreases due to service interruptions
Solution Approach 1:
The control action is localized to only those information processing apparatuses affected by liquid cooling abnormalities. Each controller independently determines whether its own apparatus is influenced by detected abnormalities and applies control only where necessary, maintaining service availability for unaffected apparatuses while ensuring reliability for affected ones.
Solution Approach 2:
Controllers continuously acquire sensor data and abnormality information, detect abnormalities, and adjust control actions based on feedback about the range of influence. This closed-loop feedback mechanism enables dynamic decision-making that balances reliability maintenance with service availability, avoiding unnecessary widespread shutdowns.
3Measurement precision
If comprehensive sensor data acquisition is implemented across all apparatuses, then measurement precision is improved for abnormality detection, but loss of information increases due to data management complexity
Solution Approach 1:
Controllers in each information processing apparatus merge sensor data from local sensors with abnormality information from other apparatuses. This combination allows comprehensive abnormality detection with high precision while distributing data management responsibilities, reducing the overall information loss through centralized data handling.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances system reliability by enabling precise identification and management of liquid cooling system abnormalities, reducing unnecessary shutdowns and improving service availability in enterprise applications.
Implementation Method 1
Heat discharged from a CPU, a memory, and the like, which are main heat generating components in the server 101 is transferred to the heat exchanger 102 by the secondary side cooling water circulated by the pump 112. The heat exchanger 102 cools the secondary side cooling water by performing heat exchange between the secondary side cooling water and the primary side cooling water circulated by pump 113.
Data Source
AI summary
There is provided an information processing apparatus among a plurality of information processing apparatuses included within an information processing system, the information processing apparatus including a memory, a processor coupled to the memory and configured to perform information processing, and a controller configured to acquire a plurality of sensor data items output from each a plurality of sensors arranged in a liquid cooling system cooling the plurality of information processing apparatuses and abnormality information relating to the liquid cooling system stored by another information processing apparatus among the plurality of information processing apparatuses, detect an abnormality of the liquid cooling system based on the acquired sensor data items and the abnormality information, obtain a range of influence of the detected abnormality of the liquid cooling system based on the detected abnormality, and control the information processing apparatus based on the range of influence.