Modular Liquid Cooling Sleds for Multi-Node Serviceability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multi-node computing systems, replacing a malfunctioning electronic component often requires taking the entire system offline due to the complexity of accessing and removing liquid cooling components like cold plates and tubing, which disrupts operations on unaffected nodes.
Innovation Solution
A modular liquid cooling system with sleds that can be slid into a chassis, allowing each sled to connect to a water circulation and power distribution network independently, enabling individual nodes to be serviced without affecting others.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Temperature
If liquid cooling components like cold plates and tubing are used to cool electronic components, then cooling effectiveness is improved, but ease of repair deteriorates because these components are difficult to access and remove
Solution Approach 1:
The system is divided into independent modular sleds, each containing electronic components and associated liquid cooling components (cold plates, tubing connections). This segmentation allows individual sleds to be removed and serviced independently without disturbing other nodes in the multi-node computing system, resolving the contradiction between maintaining effective liquid cooling and enabling easy repair of individual components.
2Reliability
If the entire multi-node computing system is taken offline to replace a malfunctioning component on one node, then reliability of component replacement is improved, but productivity deteriorates due to system downtime
Solution Approach 1:
The multi-node computing system is segmented into independent replaceable sleds that can be serviced individually. When a component on one sled malfunctions, only that specific sled needs to be removed and replaced, while other sleds continue to operate. This eliminates the need to take the entire system offline, maintaining high productivity and system availability while ensuring reliable component replacement through modular design.
3Ease of operation
If cold plates are removed to access electronic components, then ease of operation is improved for component access, but cooling effectiveness deteriorates
Solution Approach 1:
Each sled is designed as a self-contained module with electronic components mounted on a substrate that interfaces with cold plates. The entire sled assembly can be easily removed from the chassis as a single unit, providing excellent ease of operation for component access. Meanwhile, the cold plates remain securely attached to the sled, maintaining full cooling effectiveness. This modular approach eliminates the need to detach cooling components from electronic components, as both are moved together as one integrated unit.
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
Facilitates maintenance on individual nodes without disrupting the operation of other nodes, enhancing system availability and reducing downtime.
Implementation Method 1
When in the closed position, each sled is configured to (a) bridge the hot water collection network and the cold water distribution network via the corresponding cold plate
Data Source
AI summary
A computing system includes a chassis, a water circulation network coupled to the chassis, a power distribution network coupled to the chassis, and one or more sleds removably coupled to the chassis. The water circulation network includes a cold water distribution network and a hot water collection network. Each sled of the one or more sleds includes a corresponding cold plate. Each sled is configured to slide along the chassis in two slide directions including a slide-close direction to reach a closed position and a slide-open direction to vacate the closed position. When in the closed position, each sled is configured to (a) bridge the hot water collection network and the cold water distribution network via the corresponding cold plate and (b) couple to the power distribution network.