Deployment strategies

• Rolling deployment: gradually replaces old instances; watch compatibility between versions.
• Blue-green deployment: switches traffic between two environments; requires capacity and data compatibility planning.
• Canary deployment: exposes a small traffic fraction to a new version and expands after validation.
• Feature flags: separate code deployment from feature exposure; define ownership and cleanup to prevent flag accumulation.

Infrastructure principles

• Automate repeatable infrastructure changes with version-controlled configuration.
• Keep configuration separate from application code.
• Use immutable or reproducible build artifacts.
• Apply least privilege to deployment identities.
• Define readiness and liveness checks separately.
• Set resource requests, limits, and autoscaling policies based on measured behavior.
• Plan graceful shutdown, connection draining, and in-flight request handling.

Container orchestration checklist

• Health probes.
• Resource requests and limits.
• Replica count and disruption budget.
• Autoscaling signals.
• Secrets and configuration management.
• Network policies and service identity.
• Rollback and version compatibility.
• Persistent storage and backup where required.

Cloud architecture decisions

• Choose regions based on user latency, compliance, service availability, and recovery needs.
• Use managed services when their guarantees, cost, and operational model fit.
• Estimate total cost, including data transfer, storage operations, idle capacity, and observability.
• Avoid assuming that multiple replicas in one failure domain protect against a zone or region outage.