Migration Planning
Assessment, wave planning, cutover runbooks, and rollback strategy.
Tooling makes individual VM moves easy; planning is what makes a whole-estate migration safe. This guide is the methodology our migration engineers use with every enterprise customer.
Phase 1 — Assessment
- Inventory every workload: OS, sizing, storage footprint, and who owns it
- Map dependencies — which systems talk to which (the distributed firewall’s monitor mode is useful post-migration; before it, use existing monitoring or interviews)
- Classify by criticality and by migration difficulty; the easy-and-unimportant quadrant is your pilot group
- Record real utilization for right-sizing — most estates shrink 30–50% in allocated resources when sized honestly
Phase 2 — Wave Planning
Group workloads into waves of dependent systems that must move together. A typical sequence:
| Wave | Content | Purpose |
|---|---|---|
| 0 — Pilot | Low-risk internal tools | Validate process, network mapping, and team runbooks |
| 1 — Standard services | File, print, internal apps | Build routine; refine timings |
| 2 — Business applications | Line-of-business systems with their databases | The core of the migration |
| 3 — Critical systems | Revenue-critical, customer-facing workloads | Executed with the most rehearsed runbooks and tightest windows |
Phase 3 — Cutover Runbook (per wave)
- T-7 days: initial full sync complete, incremental syncs running, target network and firewall rules pre-created and tested
- T-1 day: dress rehearsal — bring replicas up in an isolated network, validate application start and data integrity, tear down
- T-0: freeze changes → final sync → start platform VMs → smoke tests → repoint DNS/load balancer → business validation sign-off
- T+1 to T+7: sources kept powered off but intact as rollback; platform backups and monitoring policies confirmed active
Rollback Strategy
Every wave needs a written rollback decision point ("if smoke tests fail by 02:00, we roll back") and a tested path: power the source back on and repoint DNS. Because sources are never modified during migration, rollback is always available until you decommission.
Our migration engineers run this process with you at no extra cost for committed migrations — from assessment through the final wave. Talk to your account team to schedule an assessment.