Beyond the traffic graph
Most teams watch link utilisation. That is necessary but not sufficient. A link can be lightly used and still be the cause of a problem.
Utilisation: watch the peaks
Five-minute averages smooth away bursts. Look at peak utilisation and the 95th percentile. If peaks regularly approach capacity, applications are already feeling it.
Latency: know your baseline
Measure round-trip time continuously. A sudden increase often means traffic has moved to a backup path or a problem has appeared somewhere along the route.
Packet loss: small numbers, big impact
TCP throughput drops sharply with even small amounts of loss, especially over longer distances. Replication that suddenly slows down is often a loss problem.
Interface errors
CRC errors and similar counters point to physical problems: dirty connectors, failing optics, damaged fibre. Catching them early prevents outages.
Connect it to the application
The most useful dashboards put network metrics beside application metrics, such as replication lag or backup duration. That makes cause and effect visible.
Share data with your provider
When you raise an issue, timestamps and metrics speed up diagnosis. Peer1IX provides per-link monitoring to support that conversation.
