Which sequence describes a practical approach to diagnosing a production ERP performance bottleneck?

• A. Reproduce issue, collect metrics, identify hotspots, profile queries, optimize SQL/indexes, review data volumes, test under load, plan deployment and rollback.
• B. Immediately scale hardware and deploy a cache without analysis.
• C. Only review frontend performance and ignore database.
• D. Increase logging verbosity without a hypothesis.

Answer: (D)

Diagnosing a production ERP performance bottleneck effectively starts with a structured, data-driven approach. Reproduce the issue in a controlled way to observe exact symptoms and gather a baseline of how the system behaves. Then collect metrics across layers—application response times, database wait statistics, query execution times, CPU, memory, disk I/O, and network latency—to quantify where the slowdown is happening. With those metrics, pinpoint hotspots—areas or operations that consume most of the time or resources. Move on to profiling the database queries involved in the slow paths to uncover slow SQL, missing indexes, suboptimal joins, or large table scans. Use those findings to optimize SQL and indexing, and also consider data volumes or data model factors that could be contributing. After applying changes, test under realistic load conditions to verify that the bottleneck is resolved and that performance remains stable under peak usage. Finally, plan the deployment with a rollback strategy so you can safely implement fixes and revert if anything goes wrong.

This approach contrasts with simply scaling hardware or adding a cache without analysis, which addresses symptoms rather than root causes; or focusing only on frontend performance and ignoring the database; or increasing logging without a guiding hypothesis, which can create noise and miss the underlying problem.