Most cloud migrations that go badly do not fail on the copy step. They fail because nobody wrote down the order of moves, what "done" means for each system, or how to get back if a cutover misbehaves. Here is a runbook skeleton you can adapt for moving on-premise servers to public, private or hybrid cloud.
1. Assessment: know what you are moving
- Inventory every server: OS, role, CPU/RAM/disk use, and who owns it.
- Map dependencies: which app talks to which database, file share, licence server or external API.
- Note hard constraints: data that must stay on-premise, latency-sensitive systems, compliance needs.
- Record a baseline for response time and backup duration so you can compare after the move.
2. Plan the waves and the rollback
- Group systems into waves, low-risk and loosely coupled first.
- For each wave, write the cutover steps, the DNS or config changes, and the owner of each step.
- Define a rollback trigger (for example, error rate or a failed smoke test) and the exact steps to point traffic back.
- Freeze changes on systems in the current wave during cutover.
3. Staged cutover
- Sync data, then run a short final sync during a planned window.
- Run smoke tests: logins, a sample transaction, scheduled jobs, backups.
- Keep the old system intact, read-only, until the new one has run cleanly for an agreed period.
4. Day-2 checks people forget
- Monitoring and alerts wired up before go-live, not after.
- Backups configured and one restore actually tested.
- Security hardening: closed ports, least-privilege access, patch schedule.
- Right-sizing review after a few weeks of real usage data.
The goal is boring cutovers. If each wave has a written plan and a tested way back, the migration stops being a single risky weekend.
More detail: assessment, staged migration and ongoing cloud management
Written by the team at Affix Center, an IT services company in Lower Parel, Mumbai.