Load Testing eKYC¶
Definition¶
Simulating production-level traffic to validate eKYC system capacity, identify bottlenecks, and ensure reliability under peak load.
Load Test Scenarios¶
| Scenario | Traffic Pattern | Purpose |
|---|---|---|
| Steady state | Constant N requests/sec | Validate baseline capacity |
| Ramp up | Gradually increase to peak | Find breaking point |
| Peak spike | Sudden 5-10x traffic burst | Validate auto-scaling response |
| Sustained peak | High traffic for hours | Memory leaks, GPU thermal throttling |
| GPU saturation | Exceed GPU capacity | Validate queue behavior, graceful degradation |
Tools¶
| Tool | Best For |
|---|---|
| Locust | Python-based, scriptable, distributed |
| k6 | JavaScript-based, developer-friendly |
| JMeter | Traditional, comprehensive |
| Gatling | Scala-based, good reporting |
Key Takeaways¶
Summary
- GPU saturation is the most likely bottleneck — test GPU scaling specifically
- Auto-scaling lag (2-5 min for new GPU instances) means you need headroom
- Test with realistic payloads — actual document/selfie images, not synthetic