CASCADING FAULT ISOLATION
Stopping it from spreading means breaking something on purpose.
01 DEFINITION
Condition in which containing a failure requires deliberately severing parts that still work, so the act of containment is indistinguishable at the surface from the failure it is containing.
02 CONDITIONS
UNCERTAINTY
CONFLICTING
The indications of the chosen break and the suffered one are the same indications.
TIME PRESSURE
SHORT
Containment only works while the spread is still ahead of it.
ERROR ASYMMETRY
ASYMMETRIC
Cutting too much and cutting too late are answered for by different people.
OPERATOR STATE
SUPERVISORY
The person is judging a spread from indications rather than seeing it.
03 WHERE IT APPEARS
ENERGYSectioning to stop a spread takes out supply that was working, and both read as loss.
TELECOMIsolating a segment to contain a fault removes service indistinguishably from the fault doing it.
MANUFACTURINGStopping upstream to protect downstream produces the same indications as the failure would.
FINANCESuspending an activity to contain exposure looks on the record like the exposure landing.
LOGISTICSHolding a hub to stop a backlog spreading is itself the thing that stops flow.
MEDICALWithdrawing a support to protect against a cascade removes a function that was working.
04 WHERE THIS IS SURFACING
This surfaces wherever parts are coupled closely enough that one can take the others. The isolating action and the fault produce the same indications, and the record afterwards carries both as outages.
What stays unresolved is how a surface tells a chosen break from a suffered one — showing them alike leaves nobody able to say what the system did on purpose, and marking them apart asks the system to be certain which it was.