Network monitoring: the handful of signals worth alerting on
KINNEX Team4 min read
A monitoring system that sends hundreds of alerts is the same as one that sends none. Choose signals that predict real trouble.
Availability
Is each device reachable, and is each uplink up? Alert on loss of a critical link or device, with a short delay to avoid noise.
Errors and quality
Rising CRC errors, discards and flapping interfaces usually point to a bad cable, a failing port or a duplex mismatch long before users complain.
Capacity
Utilisation on uplinks and the internet link, with a trend line. Congestion shows as slow applications, and trends tell you when to upgrade.
Wireless health
Client counts per access point, channel utilisation, retry rates and roaming problems.
Power and environment
PoE budget usage, UPS battery state, temperature in cabinets and server rooms.
Security-relevant events
Failed logins on network devices, configuration changes, new unknown devices and expired certificates.
Configuration
Back up device configurations automatically and flag changes. Drift is a common cause of mystery outages.
Make it actionable
Each alert should say what is wrong, where, how bad and what to do first. Review which alerts were useful each month and retire the rest.