Thermal management strategies for the modern data centre

Not too long ago, keeping a data centre ‘efficiently’ cool was the ultimate goal for Australian data centre managers. But today, with Australian data centres evolving into hubs for much larger, interconnected networks of cloud-based services - potentially servicing hundreds of thousands of global users around the clock - managing cooling loads in isolation is not enough. Customers are now looking beyond cooling at holistic thermal management solutions.

Cooling accounted for - and still accounts for - more than 35% of a data centre’s electricity bill. In a large data centre with hundreds of multikilowatt racks, shaving even 10% off the cooling bill could add up to tens of thousands of dollars in savings. But we’ve reached a point on the technology curve where data centre managers can no longer keep adding equipment to improve cooling efficiency, mainly due to space limitation and cost of real estate skyrocketing in the past few years. Connected data centres are becoming ‘containerised’ to allow for rapid deployment and decommissioning in response to equally rapid fluctuations in demand; space is as important as cost (both capital investment and operational costs).

At the same time, customers have become more sophisticated. Aside from cooling, they have spread their focus across a wider range of infrastructure priorities, including design efficiency, modularity, control, connectivity and speed. Then there are those that have taken advantage of new technologies and are opting to spread their risk across co-location facilities, centralised data centres, and both public and private cloud services. We propose a number of relatively moderate thermal management strategies to balance these factors and create a framework that delivers improved efficiencies at lower costs regardless of the size or scope of the data centre.

Proper sealing

The first strategy is a case in point. It may seem obvious but thermal gains and losses through floors, walls and ceilings, or the introduction of humidity from outside the critical facility, reduce efficiency and negatively impact other thermal management strategies.

The data centre should at all times be isolated from the general building and outside environment as much as possible. Keep doors closed and use a vapour seal to isolate the interior atmosphere. A vapour seal is one of the least expensive and most important methods for regulating the data centre’s thermal footprint and is particularly important in maintaining proper humidity levels.

While computer room precision air conditioners (CRACs) control humidity through humidification or dehumidification as required, both consume energy. An effective vapour seal - typically created using a combination of plastic film, vapour-retardant paint, vinyl wall coverings and vinyl floor systems - can reduce the amount of energy expended on humidification or dehumidification.

This is the first step in any plan to increase thermal efficiency. If the room is not properly sealed, all other measures for improving efficiency will be less effective. A data centre assessment can help identify areas where outside air is entering the controlled environment and recommend strategies for proper sealing.

Optimised airflow

Once the room is sealed, the next step is to ensure efficient air movement. The goal is to move the maximum amount of heat away from the equipment while using a minimum amount of energy.

Optimising airflow requires evaluation and optimisation of rack configuration, air-conditioner placement and cable management. Most IT equipment is designed to draw in air through the front and exhaust it out the rear. This allows equipment racks to be arranged in ‘hot aisle’ and ‘cold aisle’ configurations so that rows of racks face each other, with the front of each opposing row of racks drawing cold air from the same aisle.

This approach is most effective when cold and hot air do not mix. Therefore, perforated floor tiles should be removed from hot aisles and used only in cold aisles. Blanking panels should be used to fill open spaces in racks to prevent hot air from being drawn back through the rack.

When using the hot-aisle/cold-aisle approach, CRAC units should always be placed perpendicular to the hot aisle to reduce air travel and prevent hot air from being pulled down into the cold aisles as it returns to the air conditioner. A return ceiling plenum can also be effective in minimising the mixing of hot and cold air.

The next extension of this is containment, when hot and cold aisles are totally isolated from each other by either containing the cold aisle or the hot aisle. This strategy also gives rise to higher return temperatures to the CRAC units and gives rise to increased efficiency and capacity.

Lastly, the explosion in the number of servers that data centres must support has created cable management challenges in many facilities. If not properly managed, cables can obstruct airflow through perforated floor tiles and prevent air from being exhausted out the rear of the rack.

The underfloor plenum should be checked to determine if cabling or piping is obstructing airflow. Overhead cabling is becoming increasingly popular, which eliminates the potential for obstruction. Deeper racks are available to allow for increased airflow and existing racks can often be equipped with expansion channels to add depth for cables and airflow.

Efficient CRACs

Because of the variable nature of the IT equipment load, and the impact of other thermal management practices, CRAC units typically operate at less than 100% load all of the time.

This creates the opportunity to design systems that operate more efficiently during normal operating conditions, and since operating conditions aren’t always stable, increasing efficiency requires some method of varying capacity based on operating conditions. Technologies such as Digital Scroll compressor technology give rise to this. Digital Scroll compressor technology offers a newer way to precisely match capacity and power consumption to the desired load and can deliver significantly lower energy consumption compared to standard ‘fixed-capacity’ compressors. It allows the compressor to never be cycled off, reducing power consumption linearly as it modulates capacity, resulting in optimum system performance and control.

Supplemental cooling

The solution to rising rack densities and high room diversity is a pumped refrigerant cooling infrastructure that supports cooling modules placed directly above or alongside high-density racks to supplement the air coming up through the floor. This has a number of advantages, including increased cooling system scalability, greater flexibility and improved energy efficiency.

Higher-density applications require fluid-based cooling to effectively remove the high concentrations of heat being generated, and from an efficiency perspective, refrigerant generally performs better than water for high-density cooling.

To the future

A holistic thermal management framework goes beyond cooling to create a significant opportunity for improving efficiency in the modern data centre. In many cases, relatively simple and inexpensive changes, such as improving room sealing, moving cables or other objects that obstruct airflow or installing blanking panels, can pay immediate dividends.

In addition, new technologies, such as variable capacity room air conditioners and optimised control systems, should be considered for their impact on efficiency. Finally, supplemental cooling systems provide a response to increased equipment densities that can increase the scalability and efficiency of existing cooling systems.