# Expected Diagnosis: sustained-cpu-pressure-during-incident-window

This case should converge on a workload-driven CPU and allocation-pressure path
rather than stopping at a generic hot-query explanation.

The expected answer should make these points explicit:

- A high-frequency batch delete or framework-driven delete path becomes the
  dominant workload change during the incident window.
- A batch delete path with a large IN-list style access pattern amplifies
  allocation and GC pressure, helping explain the sustained CPU increase.
- If heap profiles are available, they should show a hotspot aligned with this
  allocation-heavy path.
- Recommendations should focus on the offending batch path, workload shaping,
  and runtime memory-pressure follow-up rather than only query latency symptoms.