Server Rack Power Delivery Control Through Local BMC Power Capping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In high-availability data centers with power oversubscription, a failure in the power delivery system can lead to circuit breaker overload, resulting in loss of power to the entire server rack, as existing power capping mechanisms rely on an external management network that may not be available during failures.
Innovation Solution
Implementing a base management controller (BMC) within each server to locally monitor power supply units and cap power consumption proportionally to the number of available units, preventing circuit breaker overload and ensuring continuous power supply.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If power oversubscription is implemented in data centers, then power efficiency and resource utilization are improved, but reliability deteriorates because circuit breaker overload can occur during power delivery failures
Solution Approach 1:
The BMC proactively monitors the operational status of power supply units and preemptively caps power consumption when failures are detected, before circuit breaker overload can occur. This preliminary action prevents the harmful effect of complete power loss while maintaining power oversubscription benefits during normal operation
Solution Approach 2:
The system continuously monitors the status of power supply units and dynamically adjusts power consumption based on real-time feedback. When PSUs fail, the BMC receives feedback about the reduced capacity and相应 caps power intake to match available capacity, preventing overload while maximizing usable power
2Device complexity
If external management network is used for power capping, then device complexity is reduced, but reliability deteriorates because the management network may be unavailable during failures
Solution Approach 1:
The BMC performs self-service by autonomously monitoring its own power supply units and independently controlling power consumption without external management network intervention. This self-contained approach ensures power capping reliability during network failures while maintaining simplicity through integrated functionality
3Productivity
If power consumption is not capped during power delivery failures, then productivity is maintained, but harmful effects increase due to circuit breaker tripping and complete power loss
Solution Approach 1:
The BMC applies preliminary anti-action by preemptively reducing power consumption when PSU failures are detected, counteracting the tendency toward circuit breaker overload. This prevents the harmful effect of complete power loss while minimizing disruption to server productivity
Data Source
AI summary
Described are techniques for preventing loss of power to a server rack due to a failure in a power delivery system. A base management controller in a server of the server rack is configured to monitor for failures in the power supply units providing power to the components of the server. Upon detecting a failure in one or more of the power supply units, the base management controller determines the number of available power supply units supplying power to the components of the server. The base management controller then caps the power drawn by the server in proportion to the number of available power supply units in the server. In this manner, circuit breakers are prevented from being triggered thereby preventing the loss of power for the entire server rack.