Power Architecture Under Stress
Redundancy is not a number you buy; it is a property you must defend at every layer, and it fails wherever two paths that were promised to be independent quietly share something. The rating on the door tells you the intended tier of protection. It does not tell you whether the two power paths share a conduit, a cooling plant, a control network, or an earthing point, and it is those shared elements, not the headline topology, that decide whether a single fault stays contained. Learn the physical chain from the substation to the rack, learn where redundancy is genuinely doubled and where it only appears to be, and you can predict how a facility will behave under stress before it is ever tested.
Knowledge check
Why can two facilities with the same nominal redundancy topology have very different real-world reliability?
Show the answer
Nominal topology describes the intended number of independent paths, but real reliability depends on whether those paths are truly independent all the way through. Hidden shared elements, a common conduit, cooling plant, control network, fuel system, or earthing point, create common-cause failure modes that defeat the topology. Two sites can both be labeled the same tier yet differ enormously depending on how ruthlessly that independence was enforced and maintained.