Operations · Monitoring
Monitoring
Collecting and evaluating signals about availability, performance and capacity so problems are detected before users report them.
FoundationalUpdated 2026-09-01
Overview
Monitoring answers known questions: is it up, is it fast enough, is it running out of space. It is built from availability checks, resource metrics, synthetic transactions and log based alerts, tied to thresholds that reflect service expectations.
The quality bar is actionability. Every alert should have an owner, a documented meaning and a response; alerts that are routinely ignored actively reduce reliability because they train people to ignore signals.
How it works
- 01Agents, SNMP, APIs and streaming telemetry collect data; a monitoring platform evaluates rules and raises alerts.
- 02Dependency aware suppression prevents one router failure generating four hundred alerts.
- 03Alert routing sends notifications to the accountable team with a defined escalation path and on call schedule.
- 04Dashboards support triage; reports support capacity and service level conversations.
Layer model
Infrastructure
Host CPU, memory, diskNetwork interfaces and errorsStorage capacity and latencyHypervisor and cluster healthPower and environmental
Platform & service
Service and process stateDatabase health and replication lagQueue depthCertificate expiryBackup job outcomes
Experience
Synthetic transactionsReal user monitoringLogin success ratesThird party and SaaS availability
Security considerations
- Monitor security relevant conditions too: logging pipelines stopping, EDR agents unhealthy, certificate expiry, backup failures.
- Protect monitoring credentials, they typically have broad read access across the estate.
Common misconfigurations
- Alerting on raw resource thresholds rather than on symptoms users experience.
- New systems never added to monitoring because it is a manual step.
- Notifications sent to a distribution list nobody owns.
- No monitoring of the monitoring platform itself.