How much time do I have until the servers start to shut down in case of a cooling system failure?
For a closed architecture cooling solution, how fast this situation might occur depends in general on the load in the rack, the thermal mass in the rack and the heat exchange with the surroundings (leaks). In extreme situations, an over temperature situation can occur within seconds. For an open architecture solution, similar factors have an impact on how fast this situation would occur. However, since the room works as a heat sink, it will typically take minutes until an over temperature is reached. Full scale tests that Liebert have done together with Hewlett Packard showed that 104ºF entering air temperature to the server (which is the temperature when most servers start to shut down) was reached after 23 minutes at 3.5kW per rack heat load in the room. At 7kW per rack heat load, it was reached after 7 minutes; and at 11kW per rack, 104ºF was reached after 2 minutes.
