Comparisons
How Rift relates to the other mock servers you might be choosing between.
- Rift vs WireMock — the JVM incumbent, and the comparison most teams evaluating Rift actually care about.
- Rift vs Microcks — the Apache-2.0, CNCF-incubating alternative. Spec-driven rather than stub-driven, and it does contract testing and seven protocols Rift does not, so the overlap is narrower than the throughput table suggests.
For Mountebank, the relationship is different in kind: Rift implements Mountebank’s admin API and loads its imposters.json unchanged, so the relevant page is Mountebank Compatibility rather than a comparison, plus the migration guide.
How we write these
Three rules, so you can calibrate how much to trust the numbers:
- Every performance claim names its hardware, date, engine versions, and the offered concurrency. Ratios move with concurrency; one without a connection count attached is not a measurement.
- The other engine gets tuned. We configure the thing we are comparing against the way its own documentation says to, publish what we changed, and publish the untuned column beside it so you can see whether the tuning mattered.
- We publish the scenarios where the comparison is not like-for-like, and say why. See the two caveats we do not bury.
The full harness lives in tests/benchmark and runs both engines, not just ours. If you think a comparison is unfair, the fastest way to show it is to re-run it — and we would rather publish a corrected number than a flattering one.