Engine failure does not halt core processing.
Reduced verdict ambiguity and faster triage. Each alarm arrives with its rule, source telemetry, enrichment, and confidence weighting — so the first question of the shift is decision, not interpretation.
Investigative continuity across the retention horizon. Schema-governed lake retention, reproducible deterministic correlation, and chain-of-custody-aligned evidence preservation are preconditions of the investigation, not deliverables to assemble under deadline.
A platform that degrades catastrophically under partial failure is a platform that fails operationally before it fails technically. Graceful degradation is the architectural commitment that core detection continues even when peripheral engines are unavailable.
Each engine is a contractual contributor, not a hard dependency. The decision arbiter records the absence of an unavailable engine in the lineage record, and the verdict is computed from the available signals under a degraded-confidence policy.
SOC operations continue through engine maintenance windows.
Before: engine outages produced detection gaps.
Operators see when degraded mode is active and which engines are missing.
Before: degradation was silent.
Verdicts in degraded mode carry adjusted confidence.
Before: confidence was uniform regardless of engine availability.
"Mendukung graceful degradation apabila salah satu engine tidak tersedia, tanpa menghentikan pemrosesan inti."
— NOGTUS Platform Specification