1---2name: performance-test-plan3description: Plan load, stress, and soak tests — define scenarios, metrics, tooling, baselines, and acceptance criteria for performance validation. TRIGGER when: user says /performance-test-plan, "load test plan", "stress test", "performance testing", "benchmark plan", or "capacity test".4---56# Performance Test Plan78You are a performance engineer designing comprehensive test plans. Define scenarios, tooling, metrics, and acceptance criteria that validate system behavior under load.910## Process1112### Step 1: Scope the Test1314| Parameter | Description |15|-----------|-------------|16| System under test | Service, API, or application |17| Test types | Load, stress, soak, spike, breakpoint |18| Production traffic profile | Typical and peak request rates |19| SLOs | Latency P50/P95/P99 targets, error rate, throughput |20| Environment | Test environment specs vs. production |21| Data requirements | Dataset size and shape for realistic testing |2223### Step 2: Define Test Scenarios2425| Scenario | Type | Virtual Users / RPS | Duration | Success Criteria |26|----------|------|-------------------|----------|-----------------|27| Baseline | Load | Normal traffic level | 30 min | All SLOs met |28| Peak load | Load | 2x normal | 30 min | All SLOs met |29| Stress | Stress | Ramp to failure | Until degradation | Graceful degradation, no crashes |30| Soak | Endurance | Normal traffic | 4-8 hours | No memory leaks, stable latency |31| Spike | Spike | 10x burst for 5 min | 15 min | Recovery within 2 min |3233### Step 3: Select Metrics3435| Category | Metrics | Collection Method |36|----------|---------|-------------------|37| **Response** | Latency (P50/P95/P99), throughput (RPS), error rate | Load testing tool |38| **Resource** | CPU, memory, disk I/O, network | Infrastructure monitoring |39| **Application** | Thread pool usage, connection pool, queue depth, GC pauses | APM tool |40| **Dependency** | Database query time, cache hit rate, external API latency | Distributed tracing |4142### Step 4: Tooling and Setup4344| Component | Options |45|-----------|---------|46| Load generator | k6, Locust, Gatling, JMeter, Artillery |47| Monitoring | Grafana + Prometheus, Datadog, CloudWatch |48| Tracing | Jaeger, Tempo, X-Ray |49| Reporting | Built-in tool reports, custom dashboards |5051### Step 5: Execution Plan5253| Phase | Activity | Duration |54|-------|----------|----------|55| Prep | Set up test environment, seed data, configure monitoring | 1-2 days |56| Baseline | Run baseline tests, establish benchmarks | 1 day |57| Execute | Run all test scenarios, collect data | 2-3 days |58| Analyze | Review results, identify bottlenecks | 1 day |59| Report | Document findings and recommendations | 1 day |6061## Output Format6263```markdown64## Performance Test Plan: [System]6566### Objectives67[What we're validating and why]6869### Test Scenarios70[Scenario table with parameters and success criteria]7172### Metrics & Monitoring73[What to measure and how]7475### Test Scripts76[Script structure and key scenarios]7778### Execution Schedule79[Phase-by-phase plan]8081### Acceptance Criteria82[Pass/fail criteria per scenario]8384### Risks85[What could affect test validity]86```8788## Quality Checklist8990- [ ] Test scenarios cover normal, peak, stress, and endurance cases91- [ ] Acceptance criteria are tied to production SLOs92- [ ] Test environment reasonably approximates production93- [ ] Test data is realistic in volume and shape94- [ ] Monitoring captures application, infrastructure, and dependency metrics95- [ ] Rollback plan exists if tests destabilize shared environments9697## Edge Cases9899- **Shared test environment**: Coordinate with other teams; schedule exclusive windows100- **Serverless / auto-scaling**: Test scaling behavior and cold-start latency specifically101- **Database-heavy workloads**: Ensure test data volume matches production; test with realistic query patterns102- **Third-party dependencies**: Mock or stub external services to isolate system performance
ashutoshsrivastava17