Reading a result
How a run is scored, and what the probe reports regardless.
One KPI is scored; everything is reported
Each probe row selects one percentile as its scoring KPI — p95 for two of the live rows, p99.9 for the third. Pass and warn thresholds apply to that percentile and to loss. Every other figure is shown anyway, so a row that passes on p95 still tells you what its tail did.
What comes back from every probe run
- p50, p95, p99 and p99.9.
- Mean, min and max.
- Packets sent, received and lost, with the loss ratio.
- Duration and achieved send rate.
Three outcomes, not two
- Pass — The scored percentile and the loss ratio are both inside the pass thresholds.
- Warn — Inside the warn thresholds but past pass — or, on the endurance row, drifted across its buckets even where the overall figure held.
- Fail — Past the warn thresholds, or the run could not be completed.
Drift
The endurance run compares its six 5-minute buckets rather than only its aggregate. An AP that starts well and degrades produces a passing overall p95 and a drifted verdict, which is the more useful of the two signals.