Data centre resilience is often discussed room by room. White space is associated with servers and racks, while grey space contains the electrical and mechanical infrastructure that keeps those assets operating. That distinction is useful for layout and ownership, but it can create gaps when teams plan power protection and cooling coverage separately.

We get a clearer design when we follow the service path from the incoming supply to the IT load. Power is received, converted, protected and distributed through grey space before it reaches rack-level equipment in white space. Heat then moves in the opposite direction, away from servers and back through the cooling system. Every handoff needs an agreed source, capacity, monitoring method and recovery sequence.

Start with the service path, not the room label

A resilient design begins with the critical load and works upstream. The typical path runs from utility or generator supply through switchgear, an uninterruptible power supply and batteries, then through downstream distribution to a rack PDU and the connected servers, storage or network equipment. The physical rooms may be separate, but the operating chain is continuous.

This approach helps us distinguish between equipment that must remain energised without interruption, equipment that can tolerate a short transfer, and equipment that can restart in a controlled sequence. It also prevents a common specification gap: providing a capable data center UPS while overlooking the distribution, controls or cooling functions required to keep protected IT equipment usable.

The same map gives consultants, electrical teams, mechanical teams and IT operators a shared reference. It makes responsibility visible at each interface and gives commissioning teams a practical basis for testing normal operation, supply loss, transfer and recovery.

Grey space carries the continuity infrastructure

Grey space commonly contains the power intake, conversion and distribution systems that support the data hall. A UPS system stabilises the supply to critical loads and bridges interruptions while an alternative source becomes available. The correct architecture depends on the critical load, phase configuration, required autonomy, redundancy plan and future capacity.

For larger installations, a three phase UPS can support substantial central loads, while a modular UPS system can allow capacity to be added in planned stages. An online UPS system provides continuous conditioned output through double-conversion operation. These are architecture choices, and selection should follow the project load profile rather than a broad room classification.

Batteries form part of the same continuity chain. Their role is immediate energy support during a supply transition or outage, so their condition cannot be treated as a routine accessory check. A battery monitoring system can track indicators such as voltage, temperature, current and impedance variation, helping operators identify developing weakness before it affects backup availability.

UPS systems for data centers should therefore be reviewed with their batteries, bypass arrangement, downstream distribution and monitoring platform. A data center power backup design is only as dependable as the components and interfaces that carry power to the protected load.

White space receives and uses protected power

White space is where IT equipment performs the processing, storage and network functions the facility exists to support. Data center racks organise this equipment while supporting cabling, access and airflow. Cabinet selection should account for equipment dimensions, weight, door configuration, cable routes and the cooling approach. These factors influence usable rack capacity and the ease of later changes.

Within the rack, the intelligent PDU is the final distribution point before power reaches IT devices. A rack PDU must match the available feed, connector arrangement, load level and redundancy design. Where the chosen model provides metering, monitoring or switching, it can also improve visibility of rack consumption and help teams manage capacity at a more useful level than the upstream total alone.

The distinction between metered, monitored and switched PDU functions should be written clearly into the specification. A metered PDU may provide local load information, a monitored PDU can support remote visibility, and a switched PDU can add outlet-level control where the selected product supports it. We should confirm the exact capability required instead of using “intelligent power distribution unit” as a catch-all description.

Cooling coverage crosses the boundary

Power protection keeps IT equipment energised, but the same equipment continues to release heat. Cooling may include room-based systems, in-row units, rear-door heat exchangers or fan-wall arrangements, depending on rack density and data hall design. Some cooling equipment may sit in white space, while pumps, controls, heat rejection equipment or electrical feeds may sit elsewhere.

The key question is which thermal functions must remain available during each operating state. The answer depends on heat load, room thermal response, available stored cooling, generator transfer time and the restart behaviour of the cooling plant. Placing every cooling component on the UPS can create a very large electrical requirement. Leaving all cooling outside the continuity plan can expose active racks to a rapid temperature rise.

We therefore define cooling coverage by sequence. Controls and selected components may need uninterrupted or short-duration support, while larger mechanical loads may transfer to generator supply. The design should state how the system behaves during utility loss, transfer, battery operation and recovery. It should also confirm that cooling restarts in an order that supports the returning IT and electrical load.

Watch the handoffs where coverage weakens

The most expensive resilience gaps often appear between disciplines. We look closely at these handoffs during design review:

  • UPS to downstream distribution: Protective devices, bypass paths, cable ratings and distribution capacity must preserve the intended redundancy through to the rack.
  • Rack PDU to IT equipment: Feed type, socket arrangement, phase balance and A/B supply design should match the actual equipment list and growth plan.
  • Cabinet to cooling system: Rack layout, blanking, cable congestion and door airflow should support the selected containment and cooling method.
  • Electrical event to cooling response: Controls, transfer logic and restart priorities should be tested as one operational sequence.
  • Monitoring to action: UPS, battery, rack power and environmental alarms need clear thresholds, ownership and escalation steps.

These interfaces also affect lifecycle cost. Oversized infrastructure can leave capacity underused, while weak coordination can lead to later distribution changes, rack rearrangement or cooling upgrades. A coordinated design gives the operations team better information for expansion and reduces avoidable disruption.

Build a coverage schedule before procurement

Before equipment is selected, we recommend creating a simple coverage schedule for every critical load. It should identify the load owner, location, normal source, alternate source, required ride-through time, redundancy level, monitoring point and restart priority. This turns broad terms such as data center power protection into testable requirements.

The schedule should also connect cabinet density with rack power and cooling capacity. A future rack plan that increases computing density can change PDU loading, upstream UPS demand and the thermal profile at the same time. Reviewing those effects together helps prevent isolated upgrades that transfer a constraint from one part of the facility to another.

For Data Center Solutions UAE projects, high ambient conditions and demanding availability expectations make this joined-up review especially valuable. We support the process by bringing grey-space power protection, white-space rack infrastructure and cooling considerations into one technical discussion. The objective is a traceable protection path from the supply source to each critical IT load, supported by a thermal strategy that remains effective during normal operation and recovery.