Data Center Job Migration via Server Chassis Fan Speed Threshold
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In data centers with blade servers, the increasing power consumption and heat generation due to compute-intensive jobs lead to higher fan speeds, increasing overall power draw and cooling demands, necessitating a more efficient way to manage thermal loads across server chassis.
Innovation Solution
Implementing a thermal-based job scheduling system that identifies compute-intensive jobs and moves their execution to servers with already active fans, thereby consolidating cooling demands and reducing the number of active fans and their speeds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If compute-intensive jobs are distributed across multiple blade servers, then processing capacity is improved, but fan speed and power consumption increase
Solution Approach 1:
The patent consolidates compute-intensive jobs onto fewer blade servers that share common cooling infrastructure, merging their thermal loads. This allows the cooling system to serve multiple workloads efficiently rather than spinning up separate cooling resources for each individual server, thereby maintaining high processing capacity while reducing overall fan power consumption.
Solution Approach 2:
The system dynamically changes the operational parameters of fan speeds based on the thermal workload distribution. By monitoring temperature and job intensity, the control system adjusts fan speeds to optimal levels - increasing them only when necessary and maintaining lower speeds when thermal loads are manageable, thus reducing unnecessary energy consumption while preserving processing capacity.
2Temperature
If fan speed is increased to cool compute-intensive jobs, then thermal management is improved, but power draw increases
Solution Approach 1:
The patent implements a feedback control mechanism where temperature sensors continuously monitor blade server thermal conditions and feed this information to the cooling control system. Based on this feedback, the system intelligently adjusts fan speeds to maintain effective cooling only when and where needed, rather than operating at high speeds continuously. This feedback loop ensures optimal cooling effectiveness while minimizing the power draw of the chassis cooling system.
3Temperature
If multiple fans operate at high speed, then cooling capacity is improved, but overall power consumption increases
Solution Approach 1:
The patent merges the cooling demands of multiple compute-intensive jobs onto shared cooling infrastructure. By consolidating workloads that generate heat onto fewer physical servers with common cooling resources, the system achieves the required total cooling capacity without needing to operate multiple fans at maximum speed simultaneously. This merging approach reduces the total energy loss while maintaining adequate cooling capacity.
Data Source
AI summary
Thermal-based job scheduling among server chassis of a data center including identifying, by a data center management module in dependence upon a threshold fan speed for each server chassis, a plurality of server chassis having servers upon which one or more compute intensive jobs are executing, the data center management module comprising a module of automated computing machinery; identifying, by the data center management module, the compute intensive jobs currently executing on the identified plurality of server chassis; and moving, by the data center management module, the execution of the compute intensive jobs to one or more servers of chassis for compute intensive jobs.