How do production support engineers systematically isolate and remediate catastrophic database query latency under heavy live traffic, and why is evaluating execution plans, table locks, and missing indexes the definitive methodology for restoring SLA compliance without triggering cascading infrastructure degradation?