How fast should a rollback be?

Under 5 minutes for the technical revert (switching traffic, redeploying previous version). The faster the rollback, the less user impact. Blue-green deployments achieve near-instant rollback. Container-based rollbacks typically take 1-3 minutes.

What if the database migration cannot be rolled back?

This is why forward-only, backward-compatible migrations are critical. Use the expand-contract pattern: new migrations add columns/tables but never remove or rename existing ones. Remove old structures in a separate, later migration after the new code is stable.

Rollback Strategies Explained

Predefined plans for safely reverting deployments — because every release needs an undo button.

Rollback Strategy

A rollback strategy is a predefined plan for reverting a software deployment to the previous stable version when issues are detected. It includes the criteria for triggering a rollback, the technical mechanism for reverting, and the procedures for communicating the rollback to stakeholders.

Explanation

Every deployment carries risk. A rollback strategy ensures the team knows exactly what to do when something goes wrong — no scrambling, no guessing, no heroes. The strategy should answer three questions: When do we rollback? (automated triggers like error rate > 5%, plus manual triggers for business-critical issues), How do we rollback? (technical mechanism), and What do we communicate? (who to notify, what to say, how to prevent recurrence). Technical rollback mechanisms vary by deployment strategy. In blue-green deployment, rollback is switching traffic back to the blue environment. In container-based deployments, rollback is deploying the previous container image tag. In serverless environments, rollback is pointing to the previous function version. Database rollbacks are the hardest — if the new version ran migrations that changed the schema, simply reverting the code may not work if the old code is incompatible with the new schema. The best rollback strategies are automated and fast. If a canary deployment detects elevated error rates, it should automatically shift traffic back to the stable version without human intervention. Manual rollbacks should still be documented in a runbook so any team member can execute them under pressure. Post-rollback, the team conducts a blameless post-mortem to understand the root cause and prevent recurrence.

Bookuvai Implementation

Every Bookuvai project has a documented rollback strategy established during the infrastructure milestone. Our CI/CD pipeline supports one-click rollback to any previous deployment. Database migrations use the expand-contract pattern so that code rollbacks never conflict with schema changes. After any rollback, our AI PM automatically creates a post-mortem ticket and schedules a review to capture learnings.

Key Facts

Automated rollback triggers (error rate, latency) are faster and more reliable than manual decisions
Database rollbacks require backward-compatible migrations (expand-contract pattern)
Post-rollback blameless post-mortems prevent recurrence

Related Terms

Frequently Asked Questions

How fast should a rollback be?: Under 5 minutes for the technical revert (switching traffic, redeploying previous version). The faster the rollback, the less user impact. Blue-green deployments achieve near-instant rollback. Container-based rollbacks typically take 1-3 minutes.
What if the database migration cannot be rolled back?: This is why forward-only, backward-compatible migrations are critical. Use the expand-contract pattern: new migrations add columns/tables but never remove or rename existing ones. Remove old structures in a separate, later migration after the new code is stable.