Ensuring Successful Operation of a Cloud Solution
What this domain tests
Operations questions focus on running workloads reliably: monitoring, logging, alerting, deployment, backup, and cost control. You should know which signals to watch and how to automate responses.
Concepts to master
- Observability. Metrics, logs, and traces answer different questions. Uptime checks probe availability from outside; alerts fire on conditions you define. Know that logs have retention and can be routed to sinks for long-term storage.
- Deployment and change. Managed instance groups support rolling updates; canary and blue/green patterns limit blast radius. Infrastructure-as-code keeps environments reproducible.
- Backup and recovery. Snapshots, scheduled backups, and cross-region copies support RPO/RTO targets. Test restores, not just backups.
- Cost and capacity. Labels enable cost attribution; committed-use and sustained-use discounts reward predictable load. Autoscaling handles variable demand.
Traps to expect
A frequent trap is confusing an alerting policy with a dashboard (dashboards display, they do not notify). Another is assuming logs are kept forever by default. Questions may also conflate health checks (routing traffic) with uptime checks (external probing). Watch for backup schedules that meet RPO but never tested for RTO.
Rule of thumb
Define the service-level objective first, then instrument the signals that measure it. Worked example: to catch a failing web tier, create an uptime check against the public endpoint plus an alert on error-rate metrics; a dashboard alone would show the failure but never page anyone. Pair this with a rolling update policy and health checks so bad instances are removed automatically.