Industry Solutions / Banking and Financial Services

Critical Power and Operational Resilience for Australian Financial Services

Financial services is the one Australian sector where the physical plant in the basement has a named regulatory owner upstairs. Somewhere in your institution there is a register of critical operations, a set of Board-approved tolerance levels, and a business continuity plan that has to survive an annual exercise. If payments, settlements or core banking depend on a data hall, a comms room or a branch network, then the switchboard, the UPS, the batteries and the CRAC units feeding those rooms are part of that plan whether anyone has written them into it or not.

The failure mode is rarely dramatic. An IMF working paper on operational resilience in digital payments documents an Australian incident in which the power supply to the data centre hosting Reserve Bank RITS infrastructure was inadvertently shut off during fire control system maintenance. The subsequent internal review identified a legacy switchboard design issue and a maintenance contractor who did not comply with procedures. No equipment failed. A planned maintenance activity took out a payments system, which is exactly the class of event a tolerance level is meant to bound and exactly the class of event that documented, competently executed maintenance prevents.

Indigi Power and Cooling is an Indigenous and Veteran owned critical power and cooling specialist, registered with Supply Nation and ICN Gateway. We work across the whole estate that financial services actually runs on, from a 1.5 kVA unit behind a regional ATM through to N+1 and 2N three-phase systems and precision cooling in a primary data hall, and we hand back commissioning records and maintenance reports written to be filed, not skimmed.

The financial services estate, by facility class

Banking power is not one problem. It is five, and they have genuinely different answers. Sizing a branch comms room like a data hall wastes capital, and sizing a data hall like a branch loses the licence conversation. The table below is the position we work from before anything is specified.

Two numbers on that table are worth dwelling on. The first is ATM draw. The machine itself is small, roughly 25 W to 100 W idle and 150 W to 300 W while dispensing or printing, but the ATM position is not just the machine. Add the topper, the illuminated surround, the through-the-wall enclosure heater or fan, and the separately powered network router, and the real protected load is often ten times the machine. We have seen fleets where the UPS carried the ATM faithfully and the site went dark anyway because the router was on an unprotected general power outlet. The second is the 10 to 15 minute runtime, which is the figure the finance industry has settled on because the UPS is not there to ride out the outage. It is there to carry the load until the generator picks up.

What APRA CPS 230 actually asks of your power and cooling

This is the part of the conversation that most power and cooling vendors avoid, and it is the part that decides whether a proposal gets through your risk function. Let us be precise about what the standard says, because precision is the whole value here.

Prudential Standard CPS 230 Operational Risk Management commenced on 1 July 2025 and applies to APRA-regulated entities: authorised deposit-taking institutions including foreign ADIs, general insurers, life companies and friendly societies, private health insurers, and registrable superannuation entity licensees. Where an entity had pre-existing contractual arrangements with a service provider, the standard's requirements applied to those arrangements from the earlier of the next renewal date or 1 July 2026. On 30 April 2026 APRA finalised targeted amendments to CPS 230 and CPG 230, effective 1 July 2026, creating limited exemptions from specific contractual requirements for categories of non-traditional service providers such as central banks, financial market exchanges, clearing and settlement facilities, regulators and government agencies. A precision cooling and UPS contractor is not in any exempt category, so if we are material to you, the full contractual requirements apply to our agreement.

Critical operations, and the sentence that captures the plant room

CPS 230 defines critical operations as processes undertaken by the entity or its service provider which, if disrupted beyond tolerance levels, would have a material adverse impact on depositors, policyholders, beneficiaries or other customers, or on the entity's role in the financial system. The standard then sets a floor. Unless it can justify otherwise, an ADI must classify payments, deposit-taking and management, custody, settlements and clearing as critical operations. An insurer must classify claims processing. An RSE licensee must classify investment management and fund administration.

Then comes the clause that matters most to anyone responsible for a plant room. For all APRA-regulated entities, the minimum list also includes customer enquiries and the systems and infrastructure needed to support critical operations. Infrastructure supporting a critical operation is itself in scope. Reinforcing this, the standard requires entities to identify and document the processes and resources needed to deliver critical operations, and the resource list it gives is people, technology, information, facilities and service providers. Facilities is named explicitly. A data hall carrying core banking is a facility supporting a critical operation, and its power and cooling are the reason it is a data hall rather than a warehouse.

Tolerance levels are a time budget, and cooling spends it fastest

For each critical operation the entity must set tolerance levels, including the maximum period of time it would tolerate a disruption to that operation. That is a time budget, and it is where power and cooling become quantitative rather than rhetorical. If your tolerance for a payments disruption is measured in minutes, then a UPS with 10 minutes of measured battery capacity and a generator that has not been load tested this year is not a control, it is an assumption.

Cooling deserves particular attention because it is routinely left off the protected load. A data hall that loses cooling but keeps IT power does not stay up. It heats until equipment throttles or trips on over-temperature, and in a well-populated modern hall that can happen in a handful of minutes, far inside most tolerance levels. If your CRAC fans, pumps and controls sit on unprotected power, your effective tolerance for a mains event is not your battery runtime, it is your thermal runtime, and nobody has measured it.

Material service providers, the register, and audit evidence

An entity must identify and maintain a register of its material service providers, where material means the entity relies on them to undertake a critical operation or they expose it to material operational risk, and that register is submitted to APRA annually. Formal agreements for material arrangements must allow APRA access to documentation, data and other information relating to the service, allow APRA to conduct an on-site visit to the service provider, and require the provider not to impede APRA in fulfilling its duties.

This is the practical consequence for a maintenance contract. If a maintenance regime is the control that keeps a critical operation inside tolerance, then the person who owns that control has to be able to produce the record. Not a summary invoice. The commissioning record, the battery impedance and load test results, the thermographic scan of the switchboard, the CRAC service report with measured supply and return temperatures, the date the transfer to bypass was last proved. That is what we mean when we say our documentation is written as audit evidence, and it is genuinely what a CPS 230 owner needs when internal audit reviews the business continuity plan or the annual continuity exercise turns up a gap.

Topology decisions that actually matter in banking

Zero transfer time for payments and core banking

An online double-conversion UPS rectifies incoming AC to DC and inverts it back to AC continuously, so the load is always fed from the inverter. When mains fails, nothing transfers. The battery simply becomes the source behind a DC bus that was already carrying the load, and the output waveform is undisturbed. That is what zero transfer time means, and it is the reason double-conversion is the only sensible answer for payments, settlements, core banking and anything doing synchronous writes to a transaction log.

Line-interactive and standby topologies do transfer, typically in the order of 2 to 10 milliseconds. Most modern switch-mode power supplies will ride through that, which is why line-interactive units are perfectly reasonable for a workstation. The trouble in a financial services context is that you are not protecting a workstation, you are protecting a chain, and the weakest hold-up time in that chain governs. A network switch with marginal hold-up, an older storage controller, a legacy card-processing appliance: any one of them dropping mid-transfer turns a clean 5 millisecond event into a reconciliation problem. In an environment where the tolerance level is written in minutes and the notification obligation for an out-of-tolerance disruption to a critical operation runs to 24 hours, that is not a trade worth making to save a few hundred dollars a unit.

N+1 in the branch, 2N in the data hall

N+1 means one more module or unit than the load requires, so any single unit can fail or be taken out for service without dropping the load. 2N means two complete independent systems, each capable of carrying the whole load, usually feeding dual-corded equipment through separate A and B paths from separate boards.

The honest position is that 2N in a branch is almost always the wrong spend. A branch comms room supports branch service, and if branch service is not itself a critical operation with a tight tolerance level, a well-maintained N or N+1 single-phase system with tested batteries and a documented service history delivers more real resilience per dollar than a redundant system nobody has load tested. Put the money into maintenance and into getting the router and the security systems onto the protected supply.

In the primary data hall the calculation inverts. Once core banking, payments or settlements are in the room, the requirement is not just redundancy but concurrent maintainability, meaning you can service any part of the electrical or mechanical system without taking load offline. That is what makes 2N and block redundant designs worth their cost. It is also what makes maintenance bypass arrangements non-negotiable. A UPS that cannot be bypassed without a shutdown forces you to choose between deferring maintenance and taking an outage, and deferring maintenance on a control that supports a critical operation is a conversation you do not want to have with internal audit.

Trading floors

Trading floors are electrically unusual. A single position with four to eight screens, a turret, a workstation and local edge kit can draw 600 W to 1.2 kW, and a floor of 60 positions is therefore a serious three-phase load sitting in what the building services drawings still call an office. Two failures are common. The first is that the trading floor was fed from general office distribution and inherits every fault and every planned shutdown on that board. The second is that heat load was estimated on an office occupancy figure and the floor runs hot, with the comms and edge racks in a cupboard that was never designed to reject that much energy. The fix is a dedicated protected sub-board with its own UPS at N+1, and supplementary precision cooling sized on measured draw for the rack positions rather than on a watts-per-square-metre rule of thumb.

ATM fleets

An ATM fleet is a logistics problem wearing an engineering costume. The individual technical answer is simple: a small online double-conversion unit, runtime sized to complete the transaction in flight, close the session cleanly and keep the machine reachable for a short grid event. The hard part is that you have hundreds of them across metropolitan, regional and remote sites, they age at different rates, and their batteries are the shortest-lived component in your entire estate. Valve-regulated lead-acid batteries in the warm, poorly ventilated cavity behind an ATM will not give you their datasheet life, and the first symptom of a flat string is usually a machine that failed to come back after an outage, discovered by a customer.

The answer is a fleet view rather than a site view: batteries tracked by installation date and site ambient conditions, replacement scheduled in geographic blocks to keep mobilisation costs sane, and remote alarm reporting so a battery fault is a work order rather than a discovery. Where an ATM is genuinely part of a critical operation, that fleet register and its test history is exactly the sort of thing that belongs alongside your continuity plan.

Precision cooling for comms rooms and regional data halls

The most common cooling defect across Australian branch networks is not an undersized unit. It is the absence of a dedicated unit at all. A comms room on the building's comfort air conditioning is cooled to suit people, on a schedule that suits people, which means it is often set back overnight and at weekends and shut down entirely during building system maintenance. Equipment does not keep office hours. A precision unit runs continuously, controls tightly on temperature and humidity, and is sized for a sensible heat ratio close to 1.0 because IT equipment produces almost no latent load, whereas a comfort unit spends part of its capacity dehumidifying air that did not need it.

The target to design against is the ASHRAE TC 9.9 recommended envelope of 18 to 27 degrees Celsius at the equipment inlet, and the important word is inlet. Measuring at the return grille or at the wall thermostat tells you about the room, not about the equipment, and rooms with poor air distribution routinely hold a comfortable average while a specific rack sits well above the envelope. In small comms rooms the usual culprits are a rack pushed against a wall with no clearance, a blanking panel plan that was never followed, and a supply grille discharging straight into a return path.

Redundancy at room level follows the same logic as power. A single split system serving a comms room that supports a critical operation is a single point of failure with no maintenance window, because servicing it means shutting it down. Two smaller units in an N+1 arrangement, rotated on a lead and lag schedule, gives you failover, lets each unit be serviced without taking the room off cooling, and evens out compressor wear. For regional halls we size at N+1 on the room's measured load and put the fans, pumps and controls on protected power.

Two compliance points apply without decoration. Electrical installation and alteration work is carried out to AS/NZS 3000, which governs the switchboard, sub-mains, protection and earthing arrangements that any UPS installation sits inside. Refrigerant handling requires ARCtick licensing under the Ozone Protection and Synthetic Greenhouse Gas Management Regulations, so any DX or split precision cooling work, including decommissioning and recovery, has to be done by a licensed technician. Where a branch or corporate site's comms room is served by the building's mechanical ventilation, AS/NZS 3666 hygiene requirements for air handling systems apply to that plant.

What Indigi does for financial services

Indicative pricing gives you a planning figure. UPS maintenance starts at $720 ex GST for a one-off visit, or $1,940 per year ex GST for a standard annual plan. Single-phase UPS installation and commissioning starts from $850 up to 3 kVA, $1,250 for 3 to 10 kVA and $1,800 for 10 to 20 kVA, which covers most ATM positions and branch comms rooms. Three-phase installation starts from $3,200 for small systems and $6,500 for medium. Cooling design and installation runs from $8,000 to $15,000 for small rooms, $25,000 to $45,000 for medium installations and $45,000 to $60,000 plus for large projects. Multi-site branch and ATM programmes are priced as a programme, because mobilisation and scheduling dominate the cost once you are past a handful of sites.

Frequently asked questions

Who does CPS 230 apply to?

CPS 230 applies to all APRA-regulated entities. That covers authorised deposit-taking institutions including banks, mutuals, credit unions, building societies and foreign ADIs, along with authorised banking non-operating holding companies. It also covers general insurers, life companies including friendly societies, private health insurers, and registrable superannuation entity licensees in respect of their business operations. For foreign ADIs, Category C insurers and eligible foreign life insurance companies, the obligations apply to Australian branch operations. The standard commenced on 1 July 2025, and for pre-existing service provider contracts the requirements applied from the earlier of the next renewal date or 1 July 2026.

What is the difference between CPS 230 and CPS 220?

CPS 220 Risk Management sets the overall risk management framework an entity must maintain across all its material risks. CPS 230 sits inside that framework and deals specifically with operational risk: managing operational risks with effective internal controls, maintaining critical operations within tolerance levels through severe disruptions, and managing the risks that come from using service providers. CPS 230 consolidated what were previously separate standards on outsourcing and business continuity management into one instrument. For a power and cooling discussion the distinction matters because CPS 230 is where the register of critical operations, the tolerance levels, the business continuity plan testing programme and the material service provider register all live.

Do ATMs work if there is no power?

Not by default. An ATM without backup power shuts down immediately when mains is lost, and any transaction in flight is left to be reconciled. Where an ATM is fitted with a UPS it will keep operating for as long as the batteries hold, which for a properly sized single-phase unit is usually long enough to complete the transaction in progress and close the session cleanly. The point most often missed is the network path. If the ATM has battery support but the router, modem or network switch feeding it sits on unprotected general power, the machine stays lit and goes offline anyway, which is a worse customer outcome than a dark screen. Protect the whole chain or the protection is cosmetic.

How much power does an ATM draw?

The machine itself typically draws around 25 W to 100 W idle and 150 W to 300 W while dispensing or printing during a transaction. Large or illuminated screens, toppers, receipt printers, cash recyclers and outdoor or through-the-wall units push the average higher. The figure that matters for sizing is the whole ATM position rather than the machine, because once you include signage, lighting, enclosure heating or ventilation and the network equipment, average daily consumption for an ATM installation with its auxiliary equipment commonly sits in the 1 kW to 3 kW range. Size the UPS on a measured reading at the position, not on the machine's nameplate.

How long do data centre UPS last?

Two different lifespans get confused here. The UPS frame and electronics generally give 10 to 15 years of service, with capacitors and fans needing preventive replacement somewhere around the 4 to 8 year mark before they fail. The batteries are much shorter lived: valve-regulated lead-acid strings typically deliver 3 to 5 years in service, sometimes up to 7 in ideal conditions, while lithium-ion systems are rated much longer and can approach the life of the frame itself. Battery life is heavily driven by ambient temperature, so a string in a warm plant room or an ATM cavity will underperform its datasheet. Runtime is a separate question again: financial services sites usually carry 10 to 15 minutes because the UPS only has to bridge to a generator.

What are the three types of UPS systems?

Standby or offline, line-interactive, and online double-conversion. A standby UPS runs the load from mains and switches to battery when it detects an outage or a severe voltage drop. A line-interactive UPS adds automatic voltage regulation, so it corrects brownouts and overvoltages without drawing on the battery, and suits small business and networking equipment. An online double-conversion UPS continuously converts incoming AC to DC and back to AC, so the load is always fed from the inverter and there is no transfer time when mains fails. For banking, payments and any environment writing transactions, online double-conversion is the appropriate choice, and it is what we specify from ATM units up to multi-megawatt data hall systems.

What should not be plugged into an UPS?

Keep high-inrush and motor loads off the protected supply. Laser printers are the classic offender, because the fuser draws a large current spike that can force the UPS to bypass or trip an output breaker. Vacuum cleaners, heaters, kettles, fridges and portable air conditioners belong on general power, and in branch environments cleaners' equipment plugged into a protected outlet is a genuinely common cause of nuisance alarms. Never daisy chain one UPS into another, and do not feed a UPS from a power board shared with heavy loads. The corollary is the more important rule in banking: things that should be on the UPS and frequently are not, including network switches, routers, modems, security and duress systems, and the cooling controls for the room itself.

Where we work

Related services and equipment

• UPS maintenance plans with reporting written to support a business continuity plan
• Battery maintenance and replacement, the shortest-lived component in a branch or ATM estate
• Single phase UPS installation and commissioning for ATM positions and branch comms rooms
• Three phase UPS installation and commissioning for trading floors and data halls
• Server room cooling Australia for comms rooms and regional halls
• CRAC unit maintenance services across DX, chilled water and condenser water plant
• Power integrity inspection for switchboard and distribution assurance
• Data centres and IT infrastructure for primary and secondary data hall design
• Secure and critical facility UPS and CRAC cooling where physical security constrains access
• Retail and commercial sector for branch networks sitting inside retail tenancies

Can you evidence your power and cooling position? Send us your site list, the operations each site supports, your current UPS and cooling arrangements and the tolerance levels you are working to. We will come back with a measured resilience position, the gaps we can see, and an installed and maintained price. Indigi Power and Cooling is Indigenous and Veteran owned, and registered with Supply Nation and ICN Gateway. Contact Indigi Power and Cooling UPS Maintenance Plans