Data Storage Thermal Management via Selective Power Interruption
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large computing facilities face operational challenges due to substantial waste heat generated by servers and storage devices, which can lead to thermal issues if not effectively managed.
Innovation Solution
Implementing a system that monitors temperature and power consumption in data storage systems, allowing for selective shutdown or cooling of specific components to maintain optimal temperatures, thereby preventing widespread shutdowns and ensuring higher availability and storage capacity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Temperature
If traditional cooling systems are used in data centers, then heat removal is achieved, but the entire facility must be cooled even when only局部 areas require cooling, leading to energy waste
Solution Approach 1:
The patent divides the data center cooling system into modular zones, each with independent temperature control. Instead of cooling the entire facility uniformly, the system segments cooling capacity to specific racks or server groups based on their thermal requirements, allowing precise temperature control where needed while avoiding energy waste in cooler areas.
Solution Approach 2:
The system implements localized cooling by directing cooling resources specifically to areas with thermal events. Temperature sensors detect hot spots in specific racks or server components, and the cooling system responds by directing airflow or activating cooling elements only in those localized areas, rather than uniformly cooling the entire data center.
2Reliability
If thermal events are not managed, then heat accumulates affecting entire facilities, but implementing comprehensive monitoring and control increases system complexity
Solution Approach 1:
The patent implements a feedback-based thermal management system where temperature sensors continuously monitor heat generation in racks and servers. When thermal events are detected, the system automatically responds by adjusting cooling resources or redistributing workloads. This closed-loop feedback mechanism ensures high reliability by preventing thermal events from propagating, while the automation reduces the perceived complexity through intelligent control algorithms.
Solution Approach 2:
The system performs preliminary thermal management by detecting temperature trends and predicting potential thermal events before they affect entire facilities. By monitoring power consumption and temperature correlations, the system takes preventive actions such as pre-cooling racks or redistributing workloads before thermal events occur, thereby maintaining high availability without requiring complex emergency response systems.
3Object-affected harmful factors
If selective component shutdown is implemented, then thermal events are contained to specific components, but this requires sophisticated temperature monitoring and control mechanisms
Solution Approach 1:
The patent extracts the thermal management control logic from centralized facility-wide systems and implements it at the rack or server level. Individual racks or server groups have their own temperature monitoring and control capabilities, allowing thermal events to be contained and managed locally. This extraction of control functions enables selective shutdown or cooling of specific components without requiring complex facility-wide control mechanisms.
Data Source
AI summary
Techniques are described for managing temperatures within a data storage system by selectively interrupting power to one or more components of the data storage system. Temperature sensors measure the temperature of various components included in racks of a data center, such as data storage modules, backplanes of data storage modules, or mass storage devices coupled to backplanes. A control device may determine that a thermal event, such as a higher than threshold temperature, is occurring in one or more components. The control device may emit signal(s) to instruct power distribution unit(s) to selectively interrupt or reduce the power sent to those component(s) exhibiting the thermal event. The components may also be instructed to reduce a number of operations being performed. In some cases, fan speeds may be selectively adjusted to cool the component(s) and thus mitigate the thermal event. Power consumption may be employed to infer the temperature of component(s).