Data Centers & Big Data Cooling Solutions

In data centers, free cooling is the practice of dissipating heat without artificially cooling air or water. Free cooling systems work by collecting air or water from the ambient environment, then circulating it into data center server rooms or individual server racks.

Free cooling is distinct from what’s known as mechanical cooling, which relies on refrigerants and compressors to cool air or liquid.
The main benefit of free cooling systems relative to mechanical cooling is simple: Free cooling uses much less energy. In turn, it can boost data center power efficiency and sustainability.

This is because mechanical cooling systems require substantial amounts of electricity to power the equipment that removes heat. Free cooling systems are much more passive, so they don't require nearly as much energy.

To operate a free cooling system, a data center must have access to air or water whose natural temperature is lower than temperatures inside the data center. As a result, free cooling is typically not viable for data centers in warm climates, during hot seasons or during warm periods of the day. Nor is it realistic for most data centers to cool servers using free cooling alone. Most facilities need mechanical cooling systems in place to sustain operations during periods when free cooling is not viable. This means that free cooling is a complement to, not a replacement for, mechanical cooling for the typical data center.

Immersion cooling uses specially formulated dielectric fluids to submerge IT hardware, enabling direct heat transfer and dramatically reducing the need for traditional air‑cooling systems. This results in higher rack densities, lower energy consumption, and significant reductions in cooling complexity.

Read more on dedicated article

Direct-to-chip (DTC) liquid cooling has emerged as one of the most efficient and scalable answers to this new era.

Read more on dedicated article

Water Usage Effectiveness (WUE) is a metric that measures how efficiently data centers use water. WUE is calculated by comparing the total water consumed by a data center to the energy consumed by its IT equipment. This means that WUE isn’t a measure of how much water a data center consumes in total. Instead, it’s a means of assessing how much water a facility uses relative to the volume of IT equipment housed in the building.

To calculate WUE, you need two data points:

1. The total water consumed by your data center, measured in liters, over a fixed period.
2. The total energy consumed by the IT equipment in your data center, measured in kilowatt-hours (kWh), over the same period.

Then, divide the first number by the second to calculate WUE.
For example, if your data center uses 100,000 liters of water per day and 50,000 kilowatt hours of energy, your WUE calculation would be:
100,000L / 50,000kWh = 2.0 L/kWh

PUE is a fundamental metric that measures the energy efficiency of dedicated data centers. Power usage effectiveness is the ratio between the total energy amount a facility consumes and the energy specifically used by the IT equipment.

IT Equipment Power: This component of PUE focuses on the power consumed by the core IT equipment within the data center, including servers, switches, storage devices, and networking infrastructure. It encompasses the energy required for data processing, computation, and transmission.

Cooling Infrastructure Power: Data centers generate substantial heat due to the operational intensity of IT equipment. To maintain optimal temperatures and prevent equipment from overheating, cooling systems are employed. The power consumed by these cooling mechanisms plays a crucial role in the overall PUE assessment.

Lighting and Miscellaneous Power: This component encompasses the power used by lighting systems, security equipment, and other miscellaneous electrical devices present in the data center.

Uninterruptible Power Supply (UPS) Losses: UPS systems provide backup power during utility outages to ensure uninterrupted operations. However, the UPS units themselves introduce inefficiencies during power conversion and conditioning processes, resulting in power losses.

Power Distribution Losses: This last component of PUE refers to the power distribution infrastructure, including transformers, switchgear, power distribution units (PDUs), and cabling. Each of these components incurs electrical resistance and associated inefficiencies, leading to power losses during the transmission of electricity from the utility source to the IT equipment.
The formula used to calculate power usage (PUE = Total Facility Energy / IT Equipment Energy) considers two factors: the total facility energy and the IT equipment energy.

Total facility power includes everything that uses power in the data center, like cooling systems, lights, and non-IT equipment. IT equipment power refers to the energy consumed by servers, storage devices, and networking gear. Aiming for a lower PUE is the way to go to save energy and make data centers greener.
The ideal PUE ratio is 1.0, as it signifies that every unit of power consumed is utilized solely by the IT equipment. In reality, however, most data centers fall within the range of 1.2 to 1.4 due to factors such as suboptimal equipment efficiency, inefficient cooling systems, and power losses in non-IT equipment.

Data centre power density is typically expressed in watts per square foot (W/sq. ft.) and indicates the concentration of power used by IT equipment in a given space. This metric helps determine how much power is necessary to support the equipment, guiding decisions about space utilisation and infrastructure design.

Data center power density is increasing due to factors like AI and HPC, as well as the rise of virtualization and cloud computing. High-density data centers can support 10 kW or more per rack, while ultra-high-density data centers can reach 85kW or higher. According to Volico, the average rack power consumption has gone from 4-5 kW a decade ago to 8-10 kW in 2020, with some data centers pushing beyond 50 kW/rack in 2022. The increased power density is driving the adoption of more efficient cooling technology.

In data center cooling systems, warm air is drawn through pads or misted areas where it absorbs moisture. As the water evaporates, it cools the air, which then improves the performance of the heat exchanger. When the outside air is cool enough, the system operates in dry mode, using only ambient air for cooling and conserving water.

RDHX or Rear Door Heat Exchangers, are structures with heat exchange coils and fans that fit on the back of an IT rack. During operation, hot server-rack airflow is forced through the RDHX device by the server fans or by additional fans installed on the device. Heat is exchanged from the hot air to circulating water from a cooler.
The air pressure drops due to the coil remain within the limits recommended by the manufacturers of servers: Air pressure drops < 25 Pa.

Data centre tiers are standardized classifications (defined by the Uptime Institute) that describe the level of redundancy, availability (uptime) and fault tolerance of a data centre.
They help set clear expectations for reliability, making it easier to choose the right infrastructure based on business needs.
Here’s a breakdown of the four main tiers:
Tier I – Basic Availability

• Infrastructure: Single path for power and cooling, no redundancy.
• Use case: Small businesses or internal systems with limited uptime needs.
• Expected yearly downtime: ~28.8 hours.

Tier II – Redundant Components

• Infrastructure: Includes redundant power and cooling components, but still has single points of failure.
• Use case: SMEs requiring moderate uptime and improved risk protection.
• Expected yearly downtime: ~22 hours.

 Tier III – Concurrently Maintainable

• Infrastructure: Dual power and cooling paths. Systems remain online during planned maintenance.
• Use case: Businesses running mission-critical applications with high availability needs.
• Expected yearly downtime: ~1.6 hours.

Tier IV – Fault Tolerant

• Infrastructure: Fully redundant systems with no single point of failure. Dual active distribution paths.
• Use case: Enterprises that demand maximum uptime — e.g., financial institutions, government, e-commerce.
• Expected yearly downtime: ~0.4 hours.

Choosing the right data centre tier depends on your uptime requirements, budget, and business criticality. Tier IV offers the highest reliability, while Tier I provides a cost-effective solution for less critical needs.