Deployment Strategy

Safe deployment patterns that minimize downtime and risk.

Context

You are planning how to deploy code. Choose strategy based on risk tolerance.

Domain Context

• Blue-Green: Two environments; switch traffic atomically
• Canary: Roll out to small % first; monitor; increase if good
• Rolling: Gradually replace instances; no downtime
• Feature Flags: Deploy code inactive; enable gradually

Instructions

1. Blue-Green: Two prod environments; instant rollback
2. Canary: Send 5% to new version; monitor errors/latency
3. Rolling: Replace 10% at a time; gradual, safe
4. Feature Flags: Deploy off, turn on for users gradually
5. Monitor: Watch error rates, latency, business metrics
6. Rollback: Quick rollback if something goes wrong
7. Incidents: Have runbook for deployment failures

Anti-Patterns

• No rollback plan; stuck if deploy goes bad
• Deploying without monitoring; don't know if it broke
• 100% traffic shift; any bug affects all users
• Manual deployments; error-prone, slow
• No canary; catch bugs by deploying to everyone

Further Reading

• Google Cloud deployment strategies
• Progressive delivery (LaunchDarkly)
• Kubernetes deployment strategies