Operations
Run the service, not just the server
Operations covers day-to-day delivery against agreed service levels. Define SLIs (measured indicators such as latency or error rate), SLOs (internal targets), and SLAs (contractual commitments with consequences). Monitor metrics, logs, and traces; alert on symptoms that matter, with thresholds and escalation paths. Baselines let you recognize abnormal behavior. Capacity management uses trends and load tests to avoid both saturation and waste; autoscaling policies need sensible minimums, maximums, and cool-down periods.
Maintenance and recovery
Patch in a planned order, test first, and keep rollback images. Backups need retention, encryption, and restore tests; replication is not a backup because it copies mistakes. Snapshots are point-in-time copies, not a complete recovery program. Incident management includes severity, roles, communication, runbooks, and blameless review. Change control records what changed, why, and how to reverse it.
Cost and inventory
Tags enable showback, chargeback, cost allocation, and ownership. Rightsizing removes idle capacity. Keep an asset and configuration inventory, review licenses, and track renewals and support contracts.
Traps
SLA versus SLO; monitoring versus observability; backup versus archive versus replication. Alert fatigue hides real incidents. A backup that has never been restored is unproven. Patching all servers at once creates a single outage.
Worked example
An application has an RTO of four hours and an RPO of fifteen minutes. Nightly backups fail the RPO. Use frequent incremental backups or continuous replication to a warm standby, then test a restore that meets four hours. Rule of thumb: design recovery from RTO and RPO backward, and prove it with a drill.