Skip to content

Service

Cloud services and migration

Cloud is not automatically cheaper, and any provider claiming otherwise has not run the numbers. What it reliably buys is resilience, remote access and an end to replacing physical servers every five years. The work is deciding what should move, what should not, and in what order.

Migration approach

Staged

never a single weekend cutover

Every stage has a scheduled window and a tested rollback before it runs.

What moves, what waits, what stays

A migration plan that treats every workload the same produces the expensive kind of project. These six categories cover most of what sits in a business of this size.

Cloud migration recommendation by workload type
WorkloadRecommendation
File sharesSharePoint or a hosted file service. Usually the easiest winMove early
Email and collaborationAlmost always already moved. Verify backup and retentionMove early
Line of business applicationsVendor support for cloud hosting varies enormouslyAssess carefully
DatabasesPerformance and licensing change materially in cloudAssess carefully
Domain controllersFull removal needs identity work first, rarely a quick moveHybrid usually
Manufacturing and control systemsLatency and vendor support generally rule this outUsually stay

Five ways cloud bills surprise people

Cloud costs are elastic in both directions, and without governance they only ever move one way.

  • Egress charges for data leaving the cloud provider, which are easy to overlook and hard to reverse
  • Virtual machines sized like the physical servers they replaced, which is nearly always oversized
  • Development and test resources left running outside working hours
  • Storage tiers chosen for performance where archival would do
  • Licensing that is cheaper on premises for some workloads, particularly database servers

Cost governance is part of the ongoing service rather than a one off exercise, because an environment nobody reviews will drift upward every quarter.

How a migration runs

Assessment first: every workload catalogued with its dependencies, its vendor support position and its performance profile. That document decides the order, and it frequently changes the plan.

Then a pilot with a small group of users on the least risky workload, which surfaces the practical problems that no assessment predicts. Then staged moves, each with a scheduled window, a communication to affected staff and a tested rollback.

Decommissioning last, and deliberately late. Leaving the old environment available and untouched for a period costs very little and has rescued more migrations than any other single practice.

Common questions

Answered before you ask.

Is cloud cheaper than running our own server?

Not automatically, and the honest answer for some workloads is no. Cloud replaces a capital purchase every five years with an operating cost every month, which changes the cash flow shape more than the total. Where cloud genuinely wins is in resilience, remote access, elasticity and not having a single physical box in a cupboard that nobody has patched since installation. Where it loses is heavy consistent compute and certain database licensing scenarios.

How disruptive is a migration?

Done in stages, far less than people expect. File shares and email typically move over a weekend with a rollback plan in place. Line of business applications need vendor involvement and a test period. The genuinely disruptive migrations are the ones attempted in one weekend as a single cutover, which is why we do not do them that way.

What is a hybrid environment?

Some workloads in the cloud and some on premises, connected so users do not notice the boundary. It is where most businesses of this size actually end up, and it is a legitimate destination rather than a halfway house. Manufacturing control systems, certain databases and equipment with vendor imposed constraints often have good reasons to stay local while everything else moves.

What happens to our server room?

It shrinks rather than disappears, in most cases. You will still have network equipment, and probably a small local presence for anything that could not move. The meaningful change is that you stop replacing physical servers on a five year cycle and stop carrying the risk of a hardware failure in a cupboard with questionable cooling.

Find out what should actually move.

The assessment includes a workload by workload review with an honest recommendation, including where staying on premises is the better answer.

CallFree assessment