Reproducible benchmarking and load testing

Design the experiment around a concrete question. A microbenchmark isolates a small operation; a load test measures a service and its dependencies. Neither alone proves production capacity. Prefer the repository's established tools, scenarios, and result format.

Choose workload and measurement model

Clarify the objective, endpoint/path, representative inputs and data size, expected traffic mix, concurrency, duration, dependencies, target environment, and acceptable latency/error thresholds. Ask before running costly, disruptive, or production-targeted load when scope or authorization is unclear.

Choose the model deliberately: