What this protects
Where your money works hardest
Cloud-first was right when the alternative was a cupboard server. Applied to every workload, it starts costing more than it saves.
Total cost of ownership
A steady workload on hardware you own is often cheaper over three years than renting the same capacity.
Flexibility where it pays
The cloud suits anything that grows, shrinks or might not be needed. Rent those, and own the rest.
The exit cost
Egress fees and migration effort belong in the business case before you sign, not in the surprise afterwards.
Cloud-first was right. The word that aged badly was \"everything\".
When the advice was written, on-premise meant a server in a cupboard with a fan that never stopped and a warranty nobody could find. Moving to the cloud was an obvious improvement, and for a great many workloads it still is.
What has changed is the cost model. Cloud prices rise, perpetual software does not, and a workload that runs flat out all day every day is often cheaper on hardware you own and stop paying for after three years.
Where cloud still wins
Anything that grows and shrinks
Seasonal demand, project teams and product launches suit a model that scales up on Friday and back down on Monday.
Email and collaboration
Microsoft 365 earns its keep. Nobody is building a better mailbox in a server room behind reception.
Anything you would rather not patch
If keeping it current is a job nobody wants, let the provider carry the maintenance burden.
Remote and hybrid teams
Staff working from three counties need services reachable from anywhere without a VPN argument.
New things with unknown demand
Before you buy hardware for a project, rent it for a quarter and find out whether it is even needed.
Anything with a compliance appetite
When a framework wants documented controls, a managed cloud platform arrives with most of them already in place.
Where it quietly stops making sense
A file server that has held the same 8TB for four years does not become cheaper by moving it. You rent the storage, you rent the compute to move the data, and you pay egress to get it out if you ever change your mind.
The workloads that hurt are the dull ones. A line-of-business application written for Windows Server 2012, a CAD file share, a workshop system that talks to a machine over a serial cable. None of them gain anything from the cloud, and several of them charge you more for the privilege.
What hybrid actually looks like
- Email, collaboration and anything new live in the cloud
- Heavy, steady, predictable workloads stay on hardware you own
- One identity across both, so staff log in once and never care where the data sits
- Backups and monitoring spanning both environments, not two systems that never speak
- A written rule for what goes where, so the next decision is not a debate
Questions to ask before you move a workload
- Does this workload actually vary, or does it run at the same level every day of the year?
- What does it cost to move the data in, and what would it cost to move it out again?
- Who is on the hook if the provider changes its pricing next April?
- Does anything need to sit close to a machine, a workshop or a person?
- What happens to the workload if the internet line drops for an afternoon?
- Is the application even supported in the cloud, or are you the first to try?
The exit cost nobody puts in the business case
Cloud agreements are easy to enter and awkward to leave. Data egress charges, re-platforming effort and the sheer time it takes to move a decade of files mean the realistic comparison is cloud versus hardware plus the cost of changing your mind.
That is not an argument against cloud. It is an argument for deciding deliberately, workload by workload, rather than applying one rule to everything because a slide deck said so in 2018.
What to do about it
- List your workloads and mark each one steady or variable. That single split decides most of it.
- Price the exit before you sign, not after. Egress and migration effort belong in the business case.
- Keep one identity across cloud and on-premise, so staff never notice which is which.
- Revisit the decision every two years. Cloud economics move faster than your hardware cycle.
Questions we get asked about this
Is hybrid just a fancy word for \"we could not decide\"?
No. It is a deliberate split: cloud for what scales and suits renting, on-premise for what runs flat out and suits owning. The deciding factor is the workload, not the fashion.
Does on-premise mean we go back to maintaining our own servers?
Not necessarily. We can manage the hardware for you under the same contract, so the only thing that changes is where it physically sits.
Can we start hybrid without a big project?
Yes. Most businesses are already hybrid without calling it that. The work is usually documentation and a clear rule for future decisions, not a wholesale move.