What a cloud migration really costs
The question always comes first, and the honest answer is that it depends on scope. What can be said is what the bill is made of and where it slips.
The migration project. Inventory, target design, cutover and verification. This is the item quotes cost, and usually the only one.
Double running during transition. The old infrastructure keeps running while the new one is set up. That overlap often lasts several months and is paid twice. It is the most systematically forgotten item.
The arrival cloud bill. It almost always starts higher than planned, because the first sizing copies the datacentre.
Operations. A cloud environment is not self-managed. Either you build the skill internally or you delegate it, but the item exists.
Three mechanisms come up every time.
Sizing copied as-is. A physical machine was sized for a three-year peak. Reproducing that size in the cloud means paying for that peak continuously, when the platform exists precisely so you no longer have to.
Non-production environments left running. Development, testing and staging have no reason to run at night and at weekends. Automatic shutdown typically saves 50 to 80% on those environments.
Data transfer. Rarely estimated before the move, often the most surprising item once measured.
A migration does not reduce costs by itself. What reduces them is what comes after: adjusting sizing to real measured load, shutting down what does not need to run continuously, replacing maintained machines with managed services where relevant, and tracking the bill month by month rather than discovering it at year end.
It is also why we advise against rewriting everything during the migration. Move first, measure second, optimise on real data.
Part of the cost of moving off a datacentre can be covered by the AWS funding programme, through a partner. It does not make the migration free, but it shifts the break-even point enough that projects judged too expensive become fundable again.
The thing to remember is sequence: the request is filed before work starts, not after.
Does a cloud migration reduce costs?
Not mechanically. A like-for-like rehost often costs more than the original datacentre. Savings come from what follows: sizing adjusted to real load, shutting down non-production environments, managed services instead of maintained machines.
How long does a migration take?
It depends on the number of applications and how coupled they are, not on infrastructure volume. An isolated application moves in weeks. A full datacentre is handled in waves over months.
Do applications need rewriting?
Not systematically. A like-for-like rehost gets you out of a datacentre quickly, with optimisation afterwards. Rewriting during the migration stacks two risks at the same time.
This maps to our AWS Migration, MAP-funded offer, or talk it through with the founder.
Let's talk about your situation.
A 30-minute call with the founder, no strings attached. Reply within 24 business hours.
No data is shared with third parties.