Cloud migration: what actually moves, what gets retired, and what it costs

Short answer: most small-business cloud migrations move less than the business expects. A good assessment usually retires more than it moves, because a fair chunk of what’s on an old server is a job Microsoft 365 already does. The bill lives in that split, not in the number of servers you have.

Cloud migration gets sold as a single decision: you’re on a server, you should be in the cloud, here’s a quote to move it. That framing is what makes the price feel arbitrary, and it’s also why so many migrations end with a business paying monthly rent on something it could have switched off. This is the version nobody sends you: what actually moves, what shouldn’t, and where the money goes.

Got a server and a date you’re worried about? Windows Server 2016 stops getting security fixes on 12 January 2027. Ring 0151 452 3060 and I’ll tell you in ten minutes whether yours is a move, a retirement or a “leave it alone for now”.

Start by finding out what the server actually does

The single most useful hour in any cloud migration happens before anyone quotes anything: somebody opens the server and writes down what’s running on it. Not what it was bought for. What it’s doing today.

Nearly every small-office server sorts into three piles.

Pile one: things Microsoft 365 already does. File shares, the shared drive everyone maps to, an old intranet, sometimes mail if it’s really old. These don’t need migrating to the cloud, they need retiring into SharePoint and OneDrive, which you’re very likely already paying for. This is usually the biggest pile and it’s the one that saves the most money, because the alternative is paying Azure rent every month for a file server that didn’t need to exist.

Pile two: things that genuinely need a server. A line-of-business application the vendor won’t host, a database, a practice or case management system with a local component, occasionally a print or licensing role that can’t move. These are the real migration. They get sized properly, costed before you decide, and moved into a cloud environment that’s been designed rather than improvised.

Pile three: things nobody uses. There’s always a pile three. A reporting tool from a system you replaced, a share for a department that no longer exists, test data from a project that ended. It gets switched off, and the only cost is the ten minutes spent checking nobody screams.

Do that split and the quote stops being arbitrary. You’re no longer paying to move a server; you’re paying to move the two things that have to move.

Lift and shift, rebuild, or replace

For whatever lands in pile two, there are three honest options and they cost very different amounts.

Lift and shift takes the server as it is and runs it in Azure more or less unchanged. It’s the fastest and the cheapest to do, and the most expensive to own, because you’re now renting a machine that’s the same size it was when somebody guessed at the spec in 2016. It’s the right answer when the application is fussy, the vendor is unhelpful, or you need the old thing gone by a date.

Rebuild means standing the application up properly in the cloud: right-sized, on managed services where they exist, with backup and security built in rather than bolted on. Costs more up front, costs noticeably less every month afterwards, and produces something a person can still understand in two years.

Replace means admitting the application is the problem and moving to something cloud-native instead. Often the best long-term answer and rarely the right one during a migration, because you’re changing how people work and where the data lives at the same time. Worth putting on the list, worth doing separately.

Most small migrations are a mix: lift and shift the awkward thing so the deadline is met, retire the file shares, and schedule the rebuild for when nothing is on fire.

What it actually costs, and the part people get wrong

There are two numbers in a cloud migration and they behave completely differently.

The project is a one-off: assessment, design, the move itself, the tidy-up afterwards. It should be a fixed price, quoted before anyone starts, with the scope written down. If a migration is quoted by the day with no scope, you’re carrying the risk of somebody else’s estimate.

The running cost is forever, and it’s the one that surprises people. Azure bills by the hour for what’s switched on, plus storage, plus data movement. A server that was a sunk cost sitting in a cupboard becomes a line item every month. That’s not a reason to avoid the cloud, but it is the reason the retire-versus-move split matters so much: every item in pile one that you move instead of retiring is a monthly bill you’ve volunteered for.

The other thing worth knowing before you sign anything: ask what the exit looks like. How do you get the data back out, in what format, and what does that cost? Most people never ask, and the answer is usually fine, but it’s a lot easier to ask before than during.

Want the split done on your server? A fixed-price assessment tells you which of your three piles is which, what each would cost to move, and what to retire instead. See Microsoft 365 and Azure services, or if you’re in the county, Azure cloud support in Cheshire.

The date that’s driving most of these conversations

Windows Server 2016 stops receiving security updates on 12 January 2027. Nothing breaks on the day. The server carries on exactly as before. What changes is that it never gets a security fix again, and it starts failing the questions your insurer, your bigger customers and any compliance questionnaire will ask.

That deadline is real, and it’s also the reason a lot of businesses are about to buy a new server that’s the same shape as the old one without ever asking what’s on it. The migration you’d have been sold is nearly always bigger than the migration you need. Finding that out costs an hour.

A short checklist before you take any quote

• Has anyone listed what’s actually running on the server, or is the quote based on the server’s existence?
• Which items are being retired into Microsoft 365 rather than moved?
• For each thing that is moving: lift and shift, rebuild, or replace, and why that one?
• What’s the one-off project price, fixed, with the scope written down?
• What’s the estimated monthly running cost afterwards, and who reviews it?
• Is the backup in the plan, and will somebody restore something to prove it works?
• How would we get our data out if we left, and what would that cost?

A provider who can answer those seven plainly is a provider who has done this before. One who can’t is quoting on hope.

Where to start

If there’s a server in a cupboard and a date in the calendar, the first move isn’t a quote for a new server and it isn’t a cloud migration. It’s an hour spent finding out what’s on the one you’ve got. Everything sensible follows from that, and quite a lot of expensive nonsense gets avoided by it.

Ring 0151 452 3060 and a person picks up. If it turns out you don’t need a migration, that’s a perfectly good outcome and I’ll say so.