How do you plan a cloud migration?

Direct answer

What a strong answer covers

How it works

A migration begins with an inventory that maps every application, its dependencies, data flows, and owners, because the surprises live in the connections. Each application is classified: rehost as is, replatform onto managed services, refactor, replace with SaaS, or retire. Security, data residency, availability, and cost requirements are defined per application before any move. A landing zone with accounts, networking, identity, and guardrails is built first. Migration proceeds in waves starting with a low-risk, self-contained application to prove the tooling and the runbook, each wave validated against its requirements, with rollback kept until validation passes and decommissioning scheduled so the old environment stops costing money.

Example

A company with 120 applications ran a six-week discovery that found 30 undocumented integrations. The first wave rehosted an internal wiki and a batch reporting job, which exposed a DNS cutover mistake in the runbook at no customer cost. The second wave replatformed the customer portal onto managed Kubernetes and a managed database with a two-week parallel run; the on-premise environment was decommissioned only after a month of clean SLO data.

Common mistakes

• Migrating the most critical system first to prove commitment.
• Lifting and shifting everything with no plan to use managed services afterward.
• Leaving the old data centre running for a year because nobody owned decommissioning.

A useful answer structure

Answer from your own work.

Do not memorise a script: use only examples you can explain honestly in follow-up questions.