Latency, response loss, and jitter
Interpret timing statistics without confusing diagnostic silence with forwarding loss.
Latency
Latency is the round-trip time of a qualifying diagnostic response, in milliseconds. The hop table includes Avg, Min, Cur, and Jit. A missing latency is unavailable rather than a measured zero.
Average and minimum latency use qualifying responses. Intermediate-hop timing and destination timing are separate measurements: Pngr sends an independent endpoint probe in each round, rather than treating the final visible router row as the only destination measurement.
ICMP, TCP, and UDP use different response mechanisms. TCP timing includes connection-call overhead, and Windows ICMP intermediate-hop timing includes .NET call overhead. Avoid treating small differences between protocols or platforms as identical packet-level measurements.
Response loss
The table's PL% and the Response loss summary count eligible diagnostic attempts without qualifying replies:
response loss (%) = 100 × (eligible attempts − qualifying replies) / eligible attemptsTimeouts and explicit network failures can count as eligible attempts. Local setup / transport errors are excluded from the eligible-attempt counter. Per-measurement details distinguish confirmed sends, attempts not sent, and cases where transmission is unknown.
Do not label diagnostic silence as confirmed forwarding loss
An intermediate router may suppress or rate-limit replies while forwarding the same traffic normally. Missing diagnostic responses are evidence about the probes, not proof that forwarded application packets were dropped.
Examples:
| Observation | Interpretation |
|---|---|
| One middle hop has high response loss, later hops and destination respond normally | Often diagnostic filtering or rate limiting; not confirmed forwarding loss |
| Destination responses disappear | Endpoint probes lack qualifying replies; investigate filtering, route, source, deadlines, and service behavior |
| Local source bind fails | Local probe setup problem; not remote packet loss |
| UDP returns destination Port Unreachable | Destination reachability established; UDP service availability not established |
Jitter
Pngr reports the mean absolute difference between adjacent qualifying latency samples. An unanswered or unavailable sample breaks the pair. Live cumulative hop jitter is unavailable until there is a qualifying adjacent pair. The rolling-window summary reports zero when it has responses but no adjacent response pair, and unavailable when it has no responses. A zero in that summary therefore does not necessarily mean enough samples established stable timing.
This is a diagnostic RTT variation measure, not an RTP media jitter measurement. The MOS est. display is an estimate based on network measurements, not a call-quality test.
Time windows and changes
Live cumulative hop counters and a selected historical time range can describe different populations. Read the active focus / range when comparing summaries. The UI also shows short and long destination averages for 30 seconds / 5 minutes.
Changing measurement configuration starts a new measurement identity for live statistics while retaining saved history. Returning to a router identity can preserve its cumulative counters, but jitter pairs do not span time spent on another router identity. Reset Statistics clears current cumulative counters while retaining saved measurements.