Performance testing · Load testing · k6

Performance and load testing: know how many users you can take

Performance and load testing measures how your system behaves as traffic grows: response times, error rates and the point where it breaks. I build k6 tests from your real traffic, run load, stress and soak scenarios against your SLOs, and tell you what to fix first and how much headroom you have.

What I test

  • Load: expected peak traffic, held long enough to see trends
  • Stress: ramp beyond peak to find the breaking point and how the system fails
  • Soak: hours at normal load to catch memory leaks and slow degradation
  • Spike: sudden bursts, such as a marketing campaign or month-end billing
  • Endpoint-level latency (p50, p95, p99) and error rates against your SLOs

How I work

  • Agree SLOs and the traffic you expect, e.g. January peak at 2× today
  • Build a traffic model from anonymised access logs, with realistic think times and data
  • Run tests against a production-sized environment, never production without agreement
  • Correlate results with your monitoring to find the real bottleneck: pool sizes, queries, CPU, caches
  • Re-test after fixes and leave a small load test running nightly as a regression check

What you get

  • A report with capacity, breaking point, bottlenecks and prioritized fixes
  • k6 scripts, thresholds and dashboards in your repository
  • A nightly regression load test in your CI

See what you get

All sample reports →

FAQ

Do you test in production?

Only if you ask for it and we agree the window and limits. By default I test a production-sized staging environment.

Which tools do you use?

k6 by default, Gatling if your team already uses it. Results go into your existing monitoring (Grafana, Datadog, CloudWatch).

Expecting a traffic peak? In 15 minutes I'll tell you how I'd find out whether you can take it.

Book a 15-min call

Prefer writing? Use the contact form. Replies within one business day, from me.