Most conversations about “the cloud” in Australia start from a false binary: either you are in a hyperscaler, or you are running a rack in a cupboard. In practice the useful question is narrower. For each workload, who needs to control it, how predictable is its demand, and where do the people using it sit? Answer those honestly and the private versus public decision mostly makes itself.
Sovereignty is about control, not a region label
Choosing an Australian region in a public cloud keeps data at rest on Australian soil, and for many organisations that is enough. It is worth being clear about what it does not do. The provider still controls the hypervisor, the management plane, the key infrastructure unless you bring your own, and the legal entity that answers to foreign process. Support staff and telemetry pipelines are frequently global. None of that is sinister, but it is not the same thing as sovereignty.
A private cloud built on open-source components changes the position. The organisation, or an Australian operator acting for it, owns the hypervisor, the storage layer and the keys. Audit logs stay inside the boundary. If a regulator, a board or a government customer asks “who can reach this data”, the answer is a short list of named people rather than a paragraph from a shared-responsibility diagram. For workloads touching health records, legal matters, defence supply chains or critical infrastructure, that shorter answer is often the whole point.
Predictable cost versus elastic cost
Public cloud pricing is elastic by design: you pay for what you consume, which is excellent when consumption is spiky or unknown. The same model becomes a liability when consumption is steady. A fleet of virtual machines that runs at the same size every day of the year is, in effect, a capital asset being rented by the hour, and egress charges add a tax on getting your own data back out.
Private cloud inverts this. Capacity is bought or leased up front, so the cost curve is flat and known, and there is no charge for moving data between your own systems. The trade-off is that you carry the risk of buying too much or too little. In our experience the organisations that benefit most are those with a stable baseline of compute that has been running for years and will run for years more. They are not paying for elasticity they never use.
Latency and where your users actually are
Australia’s geography matters more than cloud marketing admits. Round-trip time from a regional office to a Sydney public cloud region can be respectable on fibre and poor on congested NBN or satellite. For chatty applications, remote desktop sessions, voice, and anything that moves large files, tens of milliseconds are the difference between “fine” and “why is this so slow”.
A private cloud can be placed where the users are: an Australian data centre near head office, or on-site at a facility with unreliable backhaul. Placement is a design variable rather than a menu of regions. It also simplifies the network. Traffic between an office and its own cloud can stay on a private path, without hairpinning through an internet gateway and the security controls that sit on it.
When public cloud is still the right answer
None of this is an argument that public cloud is wrong. It is typically the right choice when:
- demand is genuinely unpredictable, seasonal or experimental
- you need a managed service that would be impractical to run yourself
- the workload is customer-facing and global, so proximity to overseas users matters
- the team is small and the operational burden of owning infrastructure is not worth carrying
- the data is low-sensitivity and nobody is asking the sovereignty question
Most organisations of any size end up with a mix. The mistake is not choosing one or the other; it is defaulting to one for everything without looking at the workloads individually.
A practical way to decide
A workable method is to list every significant system and score it on three axes: sensitivity of data, stability of demand, and location of users. High sensitivity, stable demand and Australian users point toward private cloud. Low sensitivity, variable demand and distributed users point toward public. The systems in the middle are where the real judgement lives, and they deserve a conversation rather than a rule.
Two further points tend to get missed. First, exit cost. Ask what it would take to leave either environment in three years, because the answer changes the calculation. Second, operations. A private cloud is only cheaper and safer if someone competent is running it around the clock, patching it, watching the storage health and testing the restores. That someone can be internal or an Australian operator, but it cannot be nobody.
Where Bizix fits
Bizix designs, builds and operates sovereign private cloud for Australian organisations, using hyperconverged compute and distributed object storage on an open-source stack managed through our own Operator UI. We also run plenty of hybrid environments, and we will say so when public cloud is the better answer for a given workload.