The cloud is fast, flexible and inexpensive — at least until we move in a system that was never designed to operate that way.

The infrastructure may now live in a more modern place. The application does not automatically become scalable, operations do not become simpler, and the cost can turn into a monthly bill that is harder to predict.

Cloud is not a good or bad destination. It is a different operating model. It creates value when the characteristics of the workload, the capabilities of the organisation and the intended business outcome all point in the same direction.

Why do we want to move to the cloud?

This question belongs before the choice of provider, region or technology.

A cloud-first strategy, an approaching hardware refresh or a management deadline does not yet explain what value the move should create.

The goal may be faster environment provisioning, support for variable demand, geographic reach, better recoverability, access to managed services or a shorter path to market for a new product. These are testable statements. They make it possible to determine whether the workload is actually better off after the migration.

Without a concrete answer to what should improve, cloud can easily become a goal instead of a means.

Lift and shift is relocation, not modernisation

A virtual machine can often be moved into a cloud environment relatively quickly. Its existing characteristics arrive with it: fixed capacity, tightly-coupled dependencies, manual deployment, legacy licensing and established operational routines.

In some situations, this can be a deliberate transitional step. It may reduce the risk of closing a data centre or create time for later modernisation. It still does not automatically use the cloud’s most important advantages.

If we rent the same oversized infrastructure and keep it running continuously, we have not built a consumption model. Only the sender of the invoice has changed.

Which workloads are at home in the cloud?

Cloud can be particularly effective where demand changes over time, environments need to appear and disappear quickly, or managed services create a meaningful operational advantage.

A workload may be a strong candidate when it:

  • needs high compute capacity periodically and much less at other times;
  • requires development, test or customer environments to be created quickly;
  • must be available across regions or from changing locations;
  • can automate deployment, scaling and recovery;
  • can reduce internal operational effort through managed databases, event services or other platform capabilities;
  • gains more business value from rapid change than from completely predictable infrastructure cost.

In these cases, cloud flexibility is not decoration. It is a response to the nature of the workload.

And which workloads can remain on the ground?

For a stable, well-understood system with consistently high utilisation, owned or dedicated infrastructure may remain economical — particularly when the equipment already exists, operations are mature and future demand is predictable.

Migration may also create less value when:

  • the application depends on latency-sensitive local systems;
  • large volumes of data must move in and out continuously;
  • licensing becomes less favourable in a cloud environment;
  • the application will soon be retired or replaced;
  • regulatory or data residency requirements significantly restrict the target;
  • the organisation cannot yet govern, secure, automate and financially track its cloud resources.

A workload is not outdated because it remains on-premises. It may simply fit its requirements best there.

Do not compare only the price of a virtual machine

The business case rarely comes down to the monthly price of a server compared with a similarly sized cloud VM.

One side includes hardware, storage, networking, energy, space, maintenance, spare capacity and operational labour. The other includes compute, storage, network traffic, support, security and observability services, licences and continuous optimisation.

Then there is the transformation itself: discovery, redesign, migration, testing, parallel operation, training and decommissioning the old environment.

A useful comparison therefore considers two complete operating models, not two price tags. Alongside cost, it weighs time, risk and the organisation’s ability to adapt.

Organisations do not migrate; workloads do

An enterprise application portfolio rarely has one correct answer.

One system should be retired. Another may be replaced by SaaS. A third can be moved with few changes because the current environment must be left quickly. A fourth needs replatforming or partial redesign to benefit from cloud services. A fifth can remain where it is.

business outcome
  → assess workload and dependencies
  → evaluate full cost and operating capability
  → choose target architecture and migration strategy
  → migrate in controlled waves
  → measure and continuously optimise

This order matters because the migration method should not be a predetermined answer. It should follow from the workload’s business and technical characteristics.

Five questions before making the decision

Five questions can quickly frame a workload’s suitability for cloud:

  1. What measurable business outcome do we expect from the migration?
  2. Do we understand its demand, data, licences and dependencies?
  3. Is relocation enough, or must the application also change?
  4. Can we operate it securely, automatically and with financial awareness?
  5. Across the full lifecycle, is the target state better than the current one?

Without answers, we do not yet have a migration plan. We only have a direction.

Do not begin with the largest system

The first migration wave should be real enough to expose missing capabilities, but bounded enough to keep learning manageable. Before it starts, record the current cost, performance, recovery time and speed of change. Measure the same outcomes again after the move.

Migration does not end at cutover. Decommissioning the old environment, tuning cloud resources and improving the decisions for the next wave are also part of the work. Each migrated workload should produce knowledge for the one that follows.

The essential point

Cloud is fast when the organisation can operate in it quickly and automatically. It is flexible when the application architecture can adapt. It is economical when consumption is visible, owned and continuously optimised.

A modern organisation is not one that runs every system in the cloud. It is one that knows why each system operates where it does.

Sources and further reading