How the measurement works
- Every few seconds a tiny request is sent to the chosen public endpoint and the round-trip time is recorded - like a ping, but over HTTPS, so values run slightly higher.
- A probe that gets no answer within 4 seconds counts as lost. Three or more lost in a row count as an outage, and its duration is measured. Each probe waits for the previous one.
- The loop runs in a web worker and the results stay in your browser. Nothing is uploaded and nothing leaves the page except the probes themselves.
Set up the test
The probe loop keeps running in a background tab; just keep the tab open. The screen-awake option needs a visible tab.
-
ms
latency now
The numbers so far
-average
-jitter
-lost probes
-uptime
-minimum
-maximum
-outages
-downtime
Timeline
under 60 ms
under 150 ms
over 150 ms
lost probe
outage
Verdict
Not enough probes yet - the verdict appears once the first ones land.
How the verdict is decided:
Stable: 0% lost, 0 outages, jitter under 15 ms, average under 150 ms.
Mostly stable: under 1% lost, 0 outages, jitter under 40 ms.
Unstable: under 5% lost and at most 2 outages.
Poor: anything worse, or downtime over 3% of the run.
Stable: 0% lost, 0 outages, jitter under 15 ms, average under 150 ms.
Mostly stable: under 1% lost, 0 outages, jitter under 40 ms.
Unstable: under 5% lost and at most 2 outages.
Poor: anything worse, or downtime over 3% of the run.
Outages
Previous runs
No finished runs yet.