Backend & systems·Resilience and failure handling
one service got slow, its callers filled up with waiting requests, and the failure walked all the way up to the front door.
Cascading failure
Failure propagating between components as each one's resources are consumed waiting on the last. It is why a non-critical dependency can cause a full outage: the criticality is in the coupling, not the feature. Timeouts, bulkheads, circuit breakers and shedding all exist to break the propagation path.