Heat Management in High-Performance Servers
The evolution of microprocessors has been a steady push toward smaller chips and faster
processing speeds. Data flows through computer circuitry as electricity, and the properties of
electricity itself cannot be easily manipulated, so processing speed is limited primarily by the
material through which the electricity flows and the distance the it has to travel. As
conduction rates increase and components shrink, electricity flows more and more quickly and
within smaller and smaller spaces.
After it has run its course, electricity dissipates as heat. The concentration of data processing
within microscopic areas on microprocessors also concentrates heat around these processing
centers, resulting in hot spots. Because the concentration of heat around sensitive components
reduces their efficiency and useful life, concomitant with the push for smaller, faster processors
has been a need for more effective ways to transfer heat away from computer processors.
Manufacturers accomplish this through heat sinks, thermal paste or grease, and cooling fans.
But the effectiveness of these internal devices can also be improved by optimizing the
environment in which they operate. Areas of stagnant, warm air around computers or servers
prevent the efficient dissipation of heat. Better circulation and cooler temperatures, on the other
hand, improve the effectiveness of internal thermal management components, increasing server
efficiency and life-expectancy. Because the thermodynamic circulation of air decreases the
amount of air that must be moved by internal system fans, these fans are allowed to do less work,
and therefore require less electricity and last longer. Manufacturers design server cabinets to
allow the free flow of air, aiding heat sinks and cooling fans in the reduction of concentrated heat
around system components. These cabinets help increase server efficiency and life-span, which
reduces costs and downtime.
See Wei, Jie. 2008. “Challenges in Cooling Design of CPU Packages for High-Performance
Servers.” Taylor and Francis Group. Heat Transfer Engineering. 29(2) pp. 178-187.
