What to check before a cloud migration becomes an outage

Migration risk is rarely caused by the visible application alone. DNS, certificates, email, background jobs, uploaded media, database access and third-party integrations all need an explicit owner and a tested path through the change.

Start with a production inventory, identify each dependency, set a measurable success condition, and write a rollback step that can be completed under pressure. A migration plan is only useful if the team can use it during an incident.

That discipline turns cloud work from a hopeful cutover into a controlled delivery operation.