Every cloud migration proposal contains a cost model. Most of them model the wrong costs.
What gets modelled
Compute, storage, egress. These are knowable and every provider has a calculator.
What does not
- Parallel running. You will run both environments for longer than planned. Budget three months minimum.
- Rework. Applications that assume a local filesystem, a fixed IP or a shared session store need changing.
- Skills. Your team needs time to learn the platform, and that time is not free.
- Egress on day one. Moving the data in is cheap. Discovering you need to move it back out is not.
A rule of thumb
Take the infrastructure estimate and add the same amount again for the first year. If that still works commercially, the migration is probably worth doing.