• Traffic is expected to grow and capacity limits are unknown.
• Latency, error rate, or saturation symptoms appear in production-like traffic.
• Team needs data-backed scaling and optimization decisions.

• Measure system behavior through Gatling simulation and throughput profiling.
• Identify bottlenecks across application, database, and dependency layers.
• Deliver capacity baseline and optimization priorities.

1. Define SLO targets and workload model (normal, peak, spike).
2. Prepare realistic traffic mix and representative test data.
3. Run baseline, ramp-up, and steady-state test phases.
4. Correlate response metrics with infrastructure signals.
5. Isolate bottleneck component and validate hypothesis.
6. Report capacity threshold, failure mode, and remediation plan.

Input:

• SLO: P95 < 300ms, peak 1200 RPS, checkout + search path

Output:

• At 1050 RPS DB pool saturation starts; P95 reaches 340ms
• CPU stable, DB wait time spikes
• Recommendation: increase pool + optimize hot query before release

Input:

• "Run a stress test"

Output (problem):

• No SLO target, no workload shape, no bottleneck evidence
• Results cannot support scale decision

• Do not run synthetic tests that ignore real traffic pattern.
• Do not conclude from average latency only.
• Do not skip dependency-level diagnostics.
• Do not compare runs with inconsistent environment baselines.
• Do not approve capacity without threshold evidence.