PCB Standoff Design for Thermal Management in Disaggregated Data Centers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face inefficiencies in energy consumption and resource management due to the need for large-scale electricity usage and the limitations of traditional hyperconverged servers, which often require simultaneous upgrades of compute, memory, and storage resources, leading to suboptimal resource utilization and increased costs.
Innovation Solution
The implementation of a data center architecture with disaggregated resources, where individual sleds house specific types of resources (compute, memory, storage) that can be logically coupled to form managed nodes, allowing for independent upgrades and improved resource allocation based on workload requirements, using a chassis-less circuit board substrate for enhanced thermal cooling and airflow.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional hyperconverged servers are used, then resource integration is achieved, but resource utilization is suboptimal and upgrades require simultaneous replacement of compute, memory, and storage
Solution Approach 1:
The server system is divided into separate resource modules (compute sleds, memory sleds, storage sleds) that can be independently upgraded and configured. Each sled type can be replaced or upgraded separately without affecting other resources, enabling flexible resource allocation and optimization based on specific workload requirements.
2Reliability
If simultaneous upgrades of compute, memory, and storage are required, then system consistency is maintained, but upgrade time and cost increase
Solution Approach 1:
By separating compute, memory, and storage into independent sleds, the system allows selective upgrading of individual components. Upgrade operations can be performed on one sled type while others continue to operate, significantly reducing upgrade time and allowing staged implementations that maintain system consistency without requiring simultaneous replacement of all resources.
3Power
If large-scale electricity usage is required for data center operations, then resource processing capability is sufficient, but energy efficiency decreases
Solution Approach 1:
The modular sled architecture enables dynamic scaling of resources based on actual workload demands. Compute, memory, and storage sleds can be added or removed independently to match processing requirements, allowing the data center to optimize its energy consumption by provisioning only the necessary resources rather than maintaining large-scale over-capacity.
4Temperature
If chassis-less circuit board substrate is used, then thermal cooling and airflow are enhanced, but structural support requirements increase
Solution Approach 1:
Mounting plates are introduced as intermediary components between the chassis-less circuit board substrate and the server chassis. These mounting plates provide the necessary structural support and mechanical attachment points while allowing the circuit board to maintain its optimized thermal design with improved airflow characteristics.
Data Source
AI summary
Systems, apparatus, articles of manufacture, and methods are disclosed to retain a printed circuit board between mounting plates. An example standoff comprises a first segment to be adjacent to a first surface of a metal plate and a second segment extending from the first segment toward a second surface of a printed circuit board (PCB), the standoff to maintain separation between the metal plate and the PCB. Also, the example standoff includes at least one of a biasing component to dampen motion between the metal plate and the PCB or a distal end on the second segment, the first segment having a first cross-sectional shape defining a first area, the distal end having a second cross-sectional shape defining a second area, the second area smaller than the first area.