Skip to content

metrics

Requests per second (RPS)

RPS measures how many requests a system or test sends each second; it is useful only when paired with latency, errors, and request mix.

Also known as: RPS, throughput

Definition

Requests per second is a throughput metric: how many requests are sent or completed in one second. It is often the headline number for API and web load tests.

In Maxoperf

Maxoperf surfaces throughput alongside latency percentiles, error facets, labels, logs, and runner health. That context matters because one RPS number can hide a failing endpoint or a saturated runner.

Common pitfalls

  • Comparing RPS across tests with different request mixes.
  • Optimizing for throughput while p95 or p99 latency is already outside the release target.

FAQ

Is higher RPS always better?

No. High RPS with unacceptable latency or errors is a failed result. Throughput must be read beside user experience and correctness.

Can VUs and RPS both be used in a load profile?

Yes. VUs describe concurrency while RPS describes request rate; many teams use both to make the intended load shape explicit.