How to make an honest cloud decision when your starting point is a tangle of legacy systems and commercial reality.
For most of the last decade, "cloud-first" has been the default answer to almost any question about technology infrastructure. It has been the right answer often enough that it has become, in many organisations, the only answer. This is a problem. Cloud-first is not a strategy. It is a default. A strategy requires the discipline to override the default when the situation warrants it.
This article is our view on what a real cloud strategy looks like, how to think honestly about which workloads belong in public cloud and which do not, and how to have the conversation with the board in a way that does not end in either evangelical enthusiasm or reflexive conservatism.
Cloud is genuinely transformative for most organisations. We are not cloud sceptics. The elasticity, the managed services, the pace at which new capabilities become available, the shift from capital to operating cost — these are real and substantial benefits that we have seen delivered in every sector we work in.
The case still has to be made, however, because the answer is workload-specific. Cloud is not a strategy, it is a deployment pattern. And like every deployment pattern, it has trade-offs. The trade-offs are not just financial — they include security architecture, regulatory compliance, operational complexity, vendor concentration and the often-underestimated cost of organisational change.
We ask three questions of every workload before recommending a migration. The order matters. Workloads that pass the first question rarely need to be asked the second or third.
This is the most important question, and it is not always answered by a financial model. A workload with significant seasonal demand, a workload that needs to scale globally, a workload that integrates with managed services unavailable on-premise — these are workloads that will be materially better in cloud. A steady-state workload with predictable demand, strong existing operational performance, and limited integration with cloud-native managed services is a workload where the answer is often "no, not materially."
This is the question that the cloud hyperscalers do not want you to ask. But it is the question that protects your organisation from vendor concentration risk. If a workload has no credible exit story — if moving it out of cloud would cost more than rebuilding it — it has created a long-term dependency that the board should know about before the migration begins.
Security architecture, regulatory compliance, latency sensitivity, data residency requirements — these are real and sometimes disqualifying constraints. A workload that processes personal information subject to Australian Privacy Principles may or may not be appropriate for a public cloud deployment; the answer depends on the controls, not on the marketing. A workload that drives plant operations may be inappropriate for any cloud deployment at all.
Cloud-first sounds decisive but it is often a way of avoiding the harder conversation about which workloads belong where. The harder conversation is the one worth having.
An honest cloud strategy is workload-specific, time-bounded and revisable. It says: these workloads will move to public cloud in this timeframe for these reasons. These workloads will move to private infrastructure for these reasons. These workloads will be retired. These workloads will remain on premise for the foreseeable future because the cost of changing them exceeds the benefit.
The strategy should also be honest about organisational change. Cloud migrations fail as often for organisational reasons as for technical ones. The internal team that runs the estate today is not necessarily the team that will run the estate tomorrow, and the skills, processes and governance arrangements need to evolve alongside the technology.
Boards have heard enough about cloud to be sceptical of evangelism. They have also heard enough about cloud security incidents to be sceptical of "lift and shift everything." The most productive conversations we have with boards about cloud are the ones that are explicit about uncertainty — about which workloads are clear candidates, which are judgement calls, and which require further analysis before a decision can be made.
The least productive conversations are the ones where the technology leader arrives with a single recommendation that the entire estate should be in cloud within three years. That recommendation almost never survives contact with the actual workload portfolio. The board knows this. The technology leader knows it too. A better conversation is one that names the workload categories, places them on a matrix, and asks the board to ratify the categorisation.
Cloud-first is not a strategy. Cloud-right is, sometimes, the right strategy. The job of an independent technology advisor is to help you tell the difference, workload by workload, without either inflating the benefits or dismissing the real transformation that is available. If you are about to have this conversation in your own organisation and would value a sounding board, please get in touch.