Interactive urges workload-first cloud planning for firms

Interactive has outlined five questions organisations should ask before making cloud or infrastructure changes, with the guidance focused on assessing workloads before choosing a platform.

Many businesses are being pushed into infrastructure decisions by ageing hardware, changes to hypervisor licensing, hardware refresh cycles, AI readiness demands and budget pressure.

Ethan Williams, Head of Presales for Cloud at Interactive, said organisations often approach the process in the wrong order, focusing on destination platforms before examining the application itself.

"You should always start with what the application is and what function it is serving for the business before you think about the platform," Williams said.

His comments reflect a broader shift in technology estates. Hybrid arrangements have become common, but often emerge through separate decisions rather than a single design plan. That can leave companies with a mix of environments that are harder to govern and more expensive to run.

Workload first

The first question is what the workload does for the business. A critical ERP system, for example, has different operational and financial implications from a development environment, analytics platform or temporary project.

That assessment includes understanding business value, criticality and dependencies, including which teams rely on the system, the likely cost of an outage and what other systems use its data. Without that groundwork, workload placement can be shaped more by supplier discussions than by operational need.

The second question is whether the workload is stable or changing. Long-running, predictable applications tend to suit environments with steady capacity and pricing, while systems under regular development may be better suited to public cloud models.

"Environments undergoing constant change can sit more easily in public cloud, whereas steady environments are much easier to run in private cloud," Williams said.

Cost pressure

The third issue is whether businesses are paying for flexibility they do not need. Public cloud pricing can make sense when organisations need experimentation, rapid provisioning or access to cloud-native services, but the economics can deteriorate when legacy applications are moved without redesign.

Brad Wells, Director of Enterprise Sales at Interactive, said lift-and-shift approaches often increase costs rather than reduce them.

"If you take a legacy application and move it into public cloud as it stands, it will be more expensive. The only reason to put it there is if you are planning to modernise it," Wells said.

Interactive also highlighted two cost areas that can be difficult to predict in advance: data egress charges and foreign exchange movements. Egress costs depend on how users interact with systems after migration, while offshore billing can change materially with movements in the Australian dollar against the US dollar.

Governance gap

The fourth question is who controls consumption and change. Cloud spending can rise steadily when server provisioning is loosely governed or configuration mistakes are left uncorrected.

Wells cited one organisation spending more than $20 million a year with a hyperscaler after costs climbed over time because staff across the business could create servers. He also described a separate case in which global replication was applied to a data transfer that was meant to use local replication, adding about $50,000 a month for four months before the problem was identified.

These examples point to governance weaknesses rather than faults with the underlying technology. The same issue can arise when optimisation recommendations remain unresolved for months, leaving environments to run at higher cost.

Williams said some organisations may be better served by buying in operational support rather than trying to build separate skills and governance structures for every platform in a hybrid estate.

The fifth question is what happens when AI-related consumption begins to scale. Assistant licences, token usage, agents and open-ended experimentation can appear manageable in the early stages, but become more significant as use spreads across a workforce.

"Everyone is being told to experiment, but the bills start arriving and the budget runs out," Williams said.

Review before move

Interactive said the answer to timing pressure is often to pause rather than rush into a migration or refresh. A broader review can identify savings of 10 to 30 per cent through changes to workload placement, procurement, licensing, commitments, asset life or tighter consumption controls.

Wells said companies should avoid responding to one infrastructure problem by creating another in a different environment.

"If you need a refresh in one part of the environment, don't react by jumping somewhere else and trading one problem for another," Wells said. "Take the time to look at the options."

Williams said customer priorities have also changed in recent years. Businesses once looked to cloud projects partly as a way to reduce headcount, but are now placing more emphasis on business value and helping existing teams work more effectively.

"It's about business value and application value," Williams said. "The teams running these environments are the glue between the business and the technology. Their role is to make sure the technology is delivering what the business actually needs."