k6

Since the previous Technology Radar edition, several k6 limitations have been resolved or significantly improved:

• All tests remain CLI-driven and test output can also be saved to JSON or CSV. These make it easier to use k6 as the part of Continuous Delivery process.
• k6 now ships with a built-in web dashboard for real-time result visualization, activated via the K6_WEB_DASHBOARD=true environment variable.
• TypeScript is supported and tests can be run directly from the k6 CLI without requiring bundlers like Webpack or Rollup.
• k6 now supports ES6+ features such as optional chaining, object spread, and private class fields making it much more aligned with the standard JavaScript ecosystem.
• The Browser testing module has graduated from experimental status to a stable core module (k6/browser), with fully async APIs. This makes it possible to simulate real user interactions in the browser on a large scale.
• Finally distributed load testing can be achieved using open-source version too thanks to the Kubernetes K6 operator.
• k6 Studio application is available to help generate test scripts using a visual interface, lowering the barrier for teams less comfortable writing tests from scratch.

k6 has been consistently developer-friendly and easy to integrate both in existing pipelines and with external monitoring services. Given the resolved limitations around TypeScript, distributed testing, and the addition of browser testing and a visual authoring tool, it has become a complete performance testing solution. We recommend it as the default choice for performance testing in future projects.

Grafana k6 is a versatile open-source performance testing tool. Test scripting is done with JavaScript, but k6 itself is written in Go for better performance. It is available as a free and open-source tool or a paid SaaS with some additional features.

k6 doesn’t offer any built-in UI for running and monitoring the tests. Instead, all the tests are executed using the CLI. This, in combination with the possibility of defining custom checks and thresholds, makes it very easy to automate k6 tests. Test output can be saved to a JSON or CSV file, or streamed to an external service such as Grafana, Prometheus, or New Relic for deeper analysis and visualization.

One of the advantages of k6 over the other performance testing tool that we used, Locust, is very detailed metrics. It measures not only the cumulative response time but also the duration of the subsequent HTTP transaction steps: e.g. the time spent handshaking a TLS session with the remote host or time to first byte.

It’s worth noting that even though tests are written in JavaScript, k6 doesn’t support Node.js modules or npm packages. You can get around it by bundling the required packages, but it’s a bit cumbersome. Another disadvantage is that distributed load testing is available only in the premium SaaS version.

k6 is developer-friendly and easy to integrate both in existing pipelines and with external monitoring services. We will definitely consider using it as a performance testing tool in future projects.