Use k6 or Oha for application load testing

Context and Problem Statement

We need lightweight tools to simulate real-world traffic for performance testing of our APIs and services.

Requirements include:

• Easy setup and usage.
• Ability to run both quick, one-off benchmarks and complex scripted scenarios.
• CI/CD integration capability.

Considered Options

• Apache JMeter
• Locust
• k6
• oha
• wrk

Decision Outcome

In reality, different tools cover different use cases.

We adopted a hybrid approach using oha for simple benchmarks and k6 for more advanced, scripted load tests.

oha is fast and easy to use for quick HTTP benchmarking (similar to wrk, but with a simpler interface and better reporting).

k6 is a developer-centric load testing tool that supports JavaScript scripting for complex test scenarios, integrates well with CI pipelines, and produces meaningful metrics.

Locust is also an acceptable choice if Python is preferred, CI-integration is not a priority, and complex user flows need to be simulated (such as authentication and other multi-step workflows).

Consequences

• ✅ oha gives us a fast, no-frills tool to test latency and throughput of HTTP endpoints.
• ✅ k6 allows scripting advanced load test scenarios using JavaScript, including auth flows and dynamic data.
• ✅ Both tools are CLI-friendly and CI-compatible.
• ❌ Splitting tools means learning two interfaces.
• ❌ k6 can be resource-intensive during heavy test scenarios.