Not everything has to go to the cloud – but a lot should
The cloud debate is often presented as a black-and-white decision: everything to the cloud or keep everything. The reality for most mid-sized companies is considerably more nuanced. There are workloads that shine in the cloud, and those that are better off staying local. The art lies in figuring out what belongs where.
A hybrid cloud strategy combines local infrastructure with cloud services – not as a lazy compromise, but as a deliberate decision to use the best of both worlds. And for many SMEs that's the most realistic and pragmatic path.
Why "everything to the cloud" is often unrealistic
There are good reasons why not everything can or should end up in the cloud. Some industry software simply doesn't run in the cloud – the application is 15 years old, the vendor no longer exists, but it works and the company depends on it. Some data must not leave the company for regulatory reasons. Some production systems need latencies in the microsecond range that can't be achieved over an internet connection.
At the same time there are workloads where on-premise no longer makes sense. Email and collaboration? Microsoft 365 is superior to the in-house Exchange server in every respect. Backup? Cloud backup is more secure, more scalable and often cheaper than local tape solutions. Dev/test environments? In the cloud you can spin them up in minutes and shut them down again in the evening.
The hybrid strategy doesn't ask "cloud or no cloud?", but "which workload belongs where?"
The decision matrix – simplified
Ask three questions for every workload: how business-critical is it? How sensitive is the data? And how easily can it be moved to the cloud?
An ERP system with highly sensitive production data and an old architecture? Stays local for now. Email and collaboration? Off to the cloud. An archive with data that's rarely needed but takes up a lot of space? Cloud storage, cheap tier. The control system for production machines? Local, but with cloud-based monitoring.
There's no universal formula. But the basic rule is: standardised services to the cloud, keep the special stuff local, and connect everything through a unified management layer.
Azure Arc and co.: the bracket for both worlds
For a long time the technical problem with hybrid cloud was that you had to manage two completely different environments – with different tools, different processes and different skill requirements. That has changed.
Microsoft Azure Arc makes it possible to manage local servers, Kubernetes clusters and even edge devices through the Azure portal – as if they were cloud resources. Policies, monitoring, security, compliance – all through one interface. AWS has a similar concept with Outposts, Google with Anthos. For SMEs that use Azure, Arc is the most natural entry point because it fits seamlessly into the existing Microsoft landscape.
That fundamentally changes the hybrid strategy. Instead of managing two islands, you have one unified system. And that makes hybrid cloud not only strategically sensible but also operationally feasible – even for smaller IT teams.
Common hybrid scenarios in mid-sized businesses
The most classic scenario – and one we at Ovarna frequently implement for manufacturing companies in the Magdeburg region: production-related systems stay local, everything else goes to the cloud. The CNC control and the MES system run on local servers with minimal latency. Office IT – email, file storage, collaboration – runs in Microsoft 365. Backup goes to the cloud. Monitoring runs through Azure Arc.
Another common scenario: the data stays local, the computing power comes from the cloud. If, for example, you want to analyse large amounts of data but don't want to invest in expensive hardware, you can run the analysis in the cloud and only bring the results back.
And then there's the classic "cloud-first, but with fallback" approach: new projects start in the cloud as a matter of principle, existing systems are migrated step by step, and for truly critical systems there's a local failover.
The path to a hybrid strategy
Don't start with the technology, but with the question: what are our business-critical processes, and what requirements do they have in terms of availability, latency, data protection and cost?
Then classify your workloads: what can go to the cloud immediately? What should be migrated in the medium term? What stays local in the long run? For each category, define the architecture and the timeline.
And then implement – step by step, not big bang. Every migration is a chance to learn, make mistakes and get better. The first wave will be bumpy. The second less so. From the third onwards it runs smoothly.
*Hybrid cloud sounds interesting? We'll develop a strategy that fits your company. Book a conversation*
