When cooling stops.
WHAT HAPPENS
Temperature rises quickly, and in a dense hall, very quickly.
WHAT THAT MEANS
Cooling failure is an emergency with a short response window.
WHAT REDUNDANCY LOOKS LIKE
More cooling capacity than needed, so one unit can fail Independent systems, where warranted
WHAT TO VERIFY
That remaining units can carry the full load.
WHY
Redundancy on paper frequently assumes conditions that do not apply on a hot day.
WHAT ELSE MATTERS
Cooling supported by generator power.
WHY
Power restored without cooling buys very little time.
WHAT TO ASK
Whether cooling is on the same backup as computing.
WHAT PLANS SHOULD EXIST
Procedures for cooling failure Thresholds at which equipment is shut down deliberately Temporary cooling arrangements
WHY DELIBERATE SHUTDOWN
An orderly shutdown is better than thermal failure.
WHAT TO TEST
Cooling failover, deliberately.
WHAT TO MONITOR
Temperature rise rate, which indicates how long you have.
