I outlined a triage-first approach: confirm scope/impact, capture AWR/ASH snapshots, check wait events and top SQL by DB time, validate recent changes (stats, plans, parameters, deployments), and review system metrics (CPU, I/O, memory, filesystem, ASM). If plan instability is suspected, compare execution plans, test plan baselines, and consider stats refresh or SQL profile. For immediate relief, throttle heavy jobs, add indexes only if justified by evidence, and communicate RCA, mitigation, and prevention steps (e.g., baselines, change control, alerting).