Measure latency as a range
A single ping hides the shape of a route. Record the minimum, median and upper percentile, then compare them over the same time window.
Network measurement, without the noise
Small observations make infrastructure easier to understand.
Northern Metric is a compact collection of practical notes about latency, name resolution and service availability.
Recent notes
Three useful measurement habits
Remember the DNS cache
A record can change instantly at its authoritative server while clients continue using a cached answer. TTL is part of the incident timeline.
Test the complete path
An open port is not the same as a working service. A useful check follows the request through name resolution, transport and the final application response.
Method
Prefer repeatable checks.
Keep the test small, timestamp the result and change one variable at a time. Reliable diagnostics are usually less dramatic—and much more useful—than a large dashboard without context.