Cloud Migration & Modernisation

Cloud migrations fail when they're sold as transformation projects. They succeed when they're sold as a series of small, reversible cutovers that each leave the business better off.

How I approach it

Discovery first — Inventory of what exists, dependency mapping, cost baseline, risk register. Two weeks. Output: a wave plan that engineering, finance, and product all agree on.

Wave-based execution — Each wave is 1–4 weeks, ends with workloads moved and the previous environment safely retired. Stateless tier first, async/batch second, stateful tier with the most care.

Cutover playbooks — Per workload: pre-checks, data sync method, traffic shift mechanism, rollback procedure, validation steps. Rehearsed in lower environments. Run with a comms cell.

Post-migration cleanup — The hidden 30% of migration cost is decommissioning. Retire old infra, cancel licenses, claw back DNS, remove shadow accounts. Done by month 3 or it never gets done.

What I won't promise

A 'big bang' migration of a multi-year-old monolith done in six weeks. The companies that win at migrations move slowly enough not to break things and quickly enough not to lose momentum.