DevOps module 07

Observe user impact, not dashboard noise.

Monitoring tells you that something is wrong. Observability helps you determine why.

Metrics

Numeric trends for utilisation, saturation, errors, latency and business outcomes.

Timestamped events with structured context, correlation identifiers and searchable fields.

End-to-end request flow across services, queues, databases and external dependencies.

Deployment, scaling, configuration and platform changes that explain behaviour.

Explicit reliability targets derived from user-visible service expectations.

Actionable notifications tied to impact, ownership, runbooks and escalation.

A technical threshold without impact may belong on a dashboard rather than a pager.

Every alert needs ownership, context and a tested first-response procedure.

Define the signal that proves service health has returned.