How to read latency and packet loss together

Ping commonly reports a test's round-trip time in milliseconds. Packet loss describes tests that did not receive their expected result within the defined deadline. A path can return quick responses for successful probes while losing others. Low latency alone therefore does not establish reliable connectivity.

Round-trip time includes both directions. It cannot by itself locate the delay on the outbound or return path. Packet size, test type, destination, sampling interval and timeout influence the result. Compare like-for-like measurements before treating a difference as a network change.

Interactive applications depend on consistency as well as speed. Two paths with similar averages can behave differently: one returns similar response times, while the other alternates between quick replies and long delays. A time series reveals changes that a single summary can hide.

In a hypothetical example, a few minutes of severe degradation surrounded by healthy hours may have a small effect on the daily average. Start with the period when the user experienced trouble, then compare the surrounding measurements. Expanding the chart's window should not erase the incident from your investigation.

A lost network test is not the same observation as an unavailable application. Some destinations limit ICMP replies; some applications fail despite responding to ping. Read web-check results alongside path measurements rather than substituting either metric for the other.

Covered lets you compare origins and domestic or international paths. Several destinations degrading from one origin gives a useful reason to investigate that origin's connectivity, without proving a cause. Keep missing measurements separate from measured loss, and avoid inventing healthy values for periods that were not observed.