Method for adaptive node positioning in chassis for efficient and utilization-based cooling adjustment
Publication Date: 2015-Nov-09
The IP.com Prior Art Database
Disclosed is a method for adaptive node positioning in server system's chassis for efficient and utilization-based cooling adjustment. The purpose is to improve cooling for rack mounted server by programmable rails.
Page 01 of 3
Method for adaptive node positioning in chassis for efficient and utilization - cooling adjustment
Servers mounted in a rack receive inlet air. Limited airflow from the floor tiles in front of each server's rack due to limited lab warehouse cooling capacity, total power consumption limitations and management, racks placement, server's density in a rack, and hot air generated by the servers. The hot air rises towards the ceiling which implies that incremental increasing hotter inlet air is provided to servers as they are placed away form the floor with the hottest inlet air being closer
to the ceiling.
Rack mounted servers are typically arranged randomly within the rack but follow some requirements such as power (PDUs capacity) and rack capacity (vertical size and stacking space between servers)
Rack mounted servers placement in a rack is not typically based on cooling needs
Cooling of servers in a rack is directly related to the power consumption based on workloads behavior and physical placement in proximity to the floor cooling vents (e.g.. systems placed away from the vents receive hotter air then those at the bottom of the rack, hence would receive hotter inlet air)
The main problem with current rack mounted servers placement is that the highest placed server receives the hottest inlet air, which could severely impact the performance and reliability of the upper servers receiving hotter inlet air
Another aspect of rack mounted servers cooling is that some servers might require colder inlet air due to cooling design limitations such as fans cooling capacity
A third aspect of cooling for such severs might be related to failure in their cooling system such as fans failure or inefficiencies
A method that could adjust cooling to each server based on its dynamic cooling requirements and needs but without increasing the lab cooling capacity is desirable
Disclosed is the idea is to improve cooling for rack mounted server by programmable rails. It provides the solution for fan failures, over heating, running high workload and basically targeted for data centers where power and cooling is a concern. This is achieved by adjusting relative position of the server horizontally on the different parameters which affect cooling any system failure. The main idea of this invention is to detect the cooling requirements for each server and dynamically adjust the inlet airflow and increase cooling to one or more servers within a rack but with the following restrictions:
Maintain current lab cooling capacity
Maintain the normal workload allocation on each server
Maintain the current thermal controls methods on each server (e.g.,Temperature sensors and Thermal management)...