Cloud that operations can live with

Lift-and-shift often moves the same operational problems into a more expensive building. The bill then records that the architecture was never finished.

Useful cloud work starts with the questions someone will ask at 2am. What failed? Who can change it? How do we get back? What did it cost to be here?

Availability, identity, cost and deployability belong in the design, not in a later workstream. AWS, Azure and Google Cloud will host a mess as willingly as they host a tidy estate.

DevOps is part of the same problem. If there is no path from commit to production, the cloud account is just another data centre with a nicer invoice.

We treat migration as a sequence with a rollback, not as a cutover weekend. Continuity first. The new platform has to be operable by the people who will still be there after the project team has left.