Performance Testing
Load Testing
Testing system behavior under expected user load to verify it meets performance requirements.
Full definition
Load testing evaluates system behavior under expected (normal) user load. It verifies that the system meets performance requirements — response times, throughput, and resource utilization — under typical production conditions.
Load testing process:
- 1.Define performance requirements (e.g., 'page loads in < 2s at 1000 concurrent users')
- 2.Create realistic user scenarios
- 3.Gradually increase load to target level
- 4.Monitor response times, error rates, and server resources
- 5.Identify bottlenecks and optimize
Key metrics to monitor:
- Response time (average, median, 95th percentile, 99th percentile)
- Throughput (requests per second)
- Error rate
- CPU, memory, disk, network utilization
- Database query times
- Connection pool usage
Popular tools:
- k6: Modern, developer-friendly, JavaScript-based
- JMeter: Mature, widely used, GUI-based
- Gatling: Scala-based, excellent for CI/CD
- Locust: Python-based, distributed testing
- Artillery: Node.js-based, YAML configuration