Migration

The decision, not the buzzword

Vendors sell the migration type they are tooled for. We don’t sell a type; we sell the choice. Each of the four below is right for some workloads and ruinous for others — the audit decides, the migration follows.

Lift & Shift

Move as-is, change nothing. Right when the driver is a deadline — a datacentre exit, a licence cliff, a contract end — and the application is healthy enough to move unopened. Wrong when the architecture is the problem: shipping a problem to new hardware ships the problem. It is the fastest type and the one most often sold when a different one was needed.

Cloud to Cloud

Provider to provider — pricing, terms, sovereignty, or a capability the current provider will never have. The trap is invisible coupling: managed services, IAM assumptions, egress paths. We map the coupling first, price the exit honestly, and design the landing so you don’t re-enter the same lock-in with a different logo.

Rearchitect

Change the structure, not just the address: the monolith that stopped scaling, the batch job that should be streams, the stateful service blocking elasticity. The most expensive type and the only one that removes limits rather than relocating them. Never undertaken on faith — the audit has to show the current architecture is the binding constraint.

Replatform

The middle path: keep the application, upgrade the platform under it — managed database instead of self-run, containers instead of VMs, a queue service instead of a hand-rolled broker. Most of the benefit of rearchitecting at a fraction of the disruption, when the application core is sound.

How the choice is made

Workload by workload, not estate by estate. A real estate migrates as a portfolio: some lift, some replatform, one or two earn a rearchitecture, and some retire — the cheapest migration of all. That portfolio, costed and sequenced, is the plan you approve before anything moves.