Cloud Migration Wave Planning Playbook
A phased program to run a large cloud migration as repeatable waves: portfolio analysis and dispositions, a migration factory, wave execution with retrospectives, and stabilize-and-close.
Cloud Migration Wave Planning Playbook
Large cloud migrations fail when treated as one giant project. The proven approach is a migration factory: a repeatable assembly line that processes applications in waves. This program builds the factory and plans the waves so hundreds of applications move predictably.
Phase-by-Phase
Portfolio Analysis. Build a complete application portfolio with dependencies, business criticality, and technical complexity. Assign each a disposition from the 7 Rs, then group applications into waves by dependency cluster and risk. Early waves should be simple to prove the factory; complex applications come later.
Migration Factory Setup. Build the factory pipeline that standardizes how an application is assessed, migrated, validated, and cut over. Define repeatable patterns for each disposition and clear acceptance gates so every wave meets the same bar.
Wave Execution. Run waves through the factory, validating and cutting over each. Capture learnings in retrospectives and feed them back so each wave is faster and smoother than the last. Velocity should rise as the factory matures.
Stabilize and Close. Stabilize the migrated estate, optimize cost now that workloads run in cloud, and decommission the source to end dual-running.
Team and Roles
A migration lead owns the wave plan and velocity. Architects own dispositions and patterns. DevOps and SRE run the factory. DBAs own data migration. Security gates each wave. Product owners accept cutovers.
Risks and Mitigations
Poor wave sequencing creates dependency deadlocks; sequence by dependency cluster. Factory bottlenecks throttle velocity; identify and widen them. Scope creep is controlled by a frozen portfolio and change control. Change fatigue is mitigated by steady cadence and visible progress.
Success Criteria
Migration velocity rises wave over wave, all in-scope applications are migrated, cost drops after optimization, and rework rate stays low.
Tooling
Terraform provisions targets, Kubernetes runs replatformed workloads, GitHub Actions powers the factory pipeline, Datadog validates each wave, and S3 stages migration data.