Reliability and performance tests
Vocabulary
Term | Description |
---|---|
VU | Stands for virtual user — simulated user running test scripts concurrently to emulate real-world traffic. |
Latency | Time between sending a request and receiving the first byte of the response. Good latency is below 300 ms. |
Throughput | Number of requests successfully handled per unit of time (e.g., requests per second). High throughput is good, low is bad. |
Iterations | How many times the test function executes (e.g., one iteration = one simulated user flow like login). |
Percentile | Percentile is more useful than average because it shows distribution. E.g., p(90) or p(95) — statistical measure indicating that X% of requests completed in Y ms or less. Example: p(90)<500 means 90% of requests completed in less than 500 ms. |
SLO (Service Level Objective) | Target performance/availability goal, e.g.: availability 99.8% uptime, performance 90% of responses within 0.5s. |
Threshold | Pass/fail condition in tool k6 tests based on metrics, used to enforce SLOs programmatically. Example: http_req_duration: ['p(95)<500']. |
Ramp-up / ramp-down | Gradual increase or decrease in the number of virtual users during a test. Ramp-up avoids sudden spikes. Ramp-down allows graceful termination. |
Sleep | Pause between actions in a test to simulate user think-time or pacing. E.g., sleep(1) means 1 second pause. |