Ask HN: Does mounting servers parallel with the temperature gradient trap heat?
Heat rises. Is heat trapped in the rack? Would mounting servers sideways (vertically) allow heat to transfer out of the rack?
Many systems have taken the vertical mount approach approach over the years: Blade servers, routers, modems, and various gaming systems.
Horizontally-mounted: parallel with the floor
Vertically-mounted: perpendicular to the floor
My guess is that vertically mounted is better, but heat flow in moving fluids is very difficult.
When there are no fans, there are some empirical rules that give an approximation of the heat flow, but are applicable to very specific cases.
If there is a cooler and a fan inside the racks, my guess is that nobody know.
Note that a big fan on the side may also help to change the air between the racks faster and increase the heat flow. Some motherboards have a fan that sucks fresh air from outside the case and send it directly to the cooler.
Probably the only way to know is to try it. If you have a few spare boards you can try the different configurations. if you measure the temperature in each case and take a few photos, it can be an interesting blog post.
Thermodynamics https://en.wikipedia.org/wiki/Thermodynamics
Are engine cylinders ever mounted horizontally? Why or why not?
> Heat rises.
Warmer air is less dense / more buoyant; so it floats.
"Does hot air really rise?" https://physics.stackexchange.com/questions/6329/does-hot-ai...
- Water ice floats because – somewhat uniquely – solid water is less dense than liquid water.
> Is heat trapped in the rack?
Probably.
> Would mounting servers sideways (vertically) allow heat to transfer out of the rack?
How could we find studies that have already tested this hypothesis?