Is cloud migration worth it: seven questions before deciding

Seven questions

Worth answering in writing, before talking to suppliers:

• What is the load profile? Variable load wins in cloud; stable, predictable load is often cheaper on-premises.
• How long can the system be down? The answer drives architecture and cost far more than the choice of provider.
• How much data can be lost? RPO sets replication and backup requirements, and those form a large part of the bill.
• Where may the data reside? In regulated sectors this constrains regions and sometimes service choice.
• What are the team’s skills? Cloud shifts work from hardware to configuration and cost control — a different skill set.
• What do costs look like in two years? Model operation rather than migration: reservations, data transfer, storage tiers.
• What happens if you need to come back? “We could not” is a legitimate answer, but it should be a deliberate one.

The most common mistake

Lift-and-shift moves virtual machines exactly as they are. It works technically, which is precisely why the mistake only surfaces with the invoice.

On-premises, running a server around the clock costs nothing extra — the hardware is already bought. In cloud it is a direct monthly cost. An architecture that was rational in your own server room becomes expensive in cloud for no benefit.

When cloud genuinely wins

Some cases are almost always yes: variable or seasonal load, a need to create and destroy environments quickly, geographically distributed users, and managed services where the team would rather not maintain infrastructure.

Conversely, stable load, strict data residency requirements and recently purchased hardware often mean a hybrid model is cheaper and simpler.