Load generator

A load generator sends a scripted load to a running service, and reports latency percentiles and errors against thresholds. It runs against a copy of the service that can take the load, never against production.

Finding it

Look for load scripts in the repository (k6 scripts, Gatling simulations, a locustfile.py) and in CI for a job, often scheduled, that runs them. That job names the target: a preview or a staging environment.

Using it

• Thresholds in the script: limits on p95 latency and the error rate, so the run fails by its exit code (thresholds in k6, assertions in Gatling).
• Run it: k6 run <script>, locust -f locustfile.py --headless -u <users> -r <rate> --run-time <duration>, or Gatling through its build plugin.
• Comparable runs: the same script, load and environment size each time, with the version under test recorded beside the result.

Examples: k6, Gatling (JVM), Locust (Python).