Treasury Corporation of Victoria

No unplanned downtime, on the platform that funds the state

Treasury Corporation of Victoria is the state’s central borrowing and investment authority. It raises the money Victoria’s agencies and departments borrow, manages their financial risk, and advises them on how to fund what they build. When TCV’s systems are unavailable, the consequences are not inconvenience.

TCV runs almost all of its IT in-house, with a small team. That combination – mission-critical workloads and a handful of people to run them – shaped every decision in this project.

The constraint was the team, not the hardware

TCV needed more capacity. That part was straightforward.

The harder requirement was that whatever replaced the existing platform had to be manageable by a small team without adding operational overhead. A more capable platform that demanded more attention would have been a step backwards, however good the specification looked on paper.

The second constraint was cost discipline. TCV’s compute and storage needs were growing at different rates, and a traditional platform forces you to buy both when you only need one. Buying compute you will not use in order to get the storage you will is a common way for public sector infrastructure budgets to disappear.

What we built

HPE Nimble Storage dHCI, which lets compute and storage scale independently, on HPE Apollo servers:

• HPE Nimble Storage dHCI, so capacity in one dimension does not force spend in the other
• HPE Apollo servers on Intel Xeon Scalable processors
• Intel Optane SSD for latency-sensitive workloads
• VMware vCenter Server as the single management plane
• VMware NSX-T for network virtualisation, with micro-segmentation enforced at the port level

The NSX-T layer is the part worth dwelling on for a financial agency. Port-level micro-segmentation means traffic between two machines in the same subnet is subject to policy rather than trusted by default, which is the difference between a breach being contained and a breach spreading. It also stretches Layer 2 networks between sites and leaves TCV able to extend into public cloud later without redesigning the network first.

Migrating a platform that cannot go down

The new infrastructure had to integrate with what TCV already ran, and the migration had to happen without interrupting operations. Not “minimal downtime” as a target. None, during business hours.

We ran weekly project meetings for the duration – not status theatre, but the mechanism for deciding what could be moved when, and what had to wait.

Onel were very knowledgeable and professional. Weekly meetings ensured the target date was met.

Veronika Mikac, Manager, Infrastructure, Treasury Corporation of Victoria

The engineer who lived through the cutover puts the important part more plainly:

No unplanned downtime during business hours. Onel identified what could and couldn’t be done safely.

Travis McKinnon, Infrastructure Engineer, Treasury Corporation of Victoria

That second sentence is the job. Anyone can list what is technically possible. Knowing what should not be attempted on a live platform, and saying so, is what you are actually buying.

What changed

Workloads run faster and availability improved, which is what a hardware refresh is expected to deliver. Three things matter more.

Management got simpler, not harder. A single vCenter plane over a dHCI platform means TCV’s small team spends less time on infrastructure, not more.

Security improved structurally. Micro-segmentation changed the network’s default posture rather than adding another product on top of it.

Spending decouples from guesswork. Compute and storage scale separately, so the next expansion buys only what is needed.

Delivered on time and on budget with no disruptions to mission critical operations.

Warren Murray, General Manager Information Technology, Treasury Corporation of Victoria

Better security, simpler management, more powerful performance and reliable data recovery.

Veronika Mikac, Treasury Corporation of Victoria