Description

In the upper summary the field Request duration seems to be calculated by including concurrency into the calculations. That is problematic when looking at durations/latencies.

Unlike throughput ( requests / s the field about) where a higher concurrency is directly related to the result for latency, this can't be calculated this way - latency remains largely unaffected by concurrency. Or in other words (ignoring server limits) doubling the concurrency doubles the throughput but the latency remains the same.

This is in so far problematic that it gives the misleading impression that higher concurrency results in lower latencies, if these numbers are used to measure SLA's / SLO's they won't reflect the latency/duration of requests in a realistic way.

Expected Behaviour

The Request duraction to be calculated based on the latency of the requests not as total time / (requests/s). Reflecting the numbers in the detailed stats below.

Actual Behaviour

The Request duraction is calculated with the assumption that concurrency linearly impacts latency.

Affected Version

latest

Steps to Reproduce

1. run a test w/ concurrency = 1
2. run a test w/ concurrency = 2
3. compare Request duration it will have approximately halved while in reality, the duration of a request has not changed significantly

Checklist

• I have read the contributing guidelines
• I have verified this does not duplicate an existing issue