Data centers produce large volumes of ambient heat, driven primarily by two factors. First, server hardware—including central processing units (CPUs) and graphics processing units (GPUs)—continually process and transmit data. These physical properties result in internal electrical resistance, emitting heat as a byproduct. Second, data centers consume substantial quantities of electricity. Added to these factors are inherent inefficiencies related to power conversion, which can generate ambient heat and raise a data center’s heat load. To maintain safe operation, data centers rely on heat sinks to eliminate this thermal surplus.
Air-cooled heat sinks have long functioned as the default heat management system in data centers. While they offer some advantages, namely relative simplicity and cost benefits, they also present significant performance challenges. Air-cooled systems have limited capacity, considerable space requirements, and are strongly dependent on the ambient temperature. They also consume huge amounts of energy, making heat loads more intense and contributing to a cycle of excess ambient heat buildup.
To address the downsides of air-cooled systems, engineers have developed liquid cooling for data centers. Fluids offer up to 1,200 times the heat-conducting capacity of air, allowing them to confer enormous efficiency gains. Fluid-based heat sinks also provide better reliability, more flexibility, and the potential to avoid complex site engineering and retrofitting.
This guide explores air vs. liquid cooling in data centers, examining the pros and cons of each option. Moreover, it profiles typical use cases for the different technologies and offers insights into which system works better for specific sites and conditions.
Types of Heat Sinks
Data centers most frequently use one of three main types of heat sinks: active, passive, and liquid-based systems. Each offers its own distinct set of advantages and drawbacks, with liquid heat management strategies delivering a particularly appealing combination of practicality and performance.
Active Heat Sinks
Active heat sinks use physical infrastructure—for instance, fans, heat pipes, and thermoelectric coolers—to dissipate heat. They are referred to as “active” systems because they employ proactive, mechanical force to remove the surrounding heat.
Forced air-cooling systems rank among the most common types of active heat sinks. These setups leverage a network of fans and blowers to direct heated air into a sink mechanism, removing it from the facility.
Pros and Cons
Active systems can provide effective thermal management control, especially in dense environments and in data centers with high-performance equipment. This is beneficial for:
- Reliability and simplicity
- Reasonable cost profiles
- Ready adaptability and integration into data center infrastructure
Nonetheless, active systems also come with several notable limitations, including:
- Restricted cooling capacity in dense environments and data centers with particularly high heat loads
- Dust and particulate matter buildups that contribute to high maintenance needs
- Noisy, fan-based systems with high electricity input requirements
Passive Heat Sinks
Passive systems chiefly rely on natural processes such as convection and conduction to draw heat out of the environment. They directly absorb the heat generated by computer hardware and other components, using the principles of thermal conductivity to transfer this heat onto a surface for dissipation. Such surfaces typically comprise either finned or pin-based designs. To illustrate, finned systems use multiple smaller and thinner flat surfaces as heat sinks, while pin-based systems employ rod shapes to increase the available surface area.
Pros and Cons
Passive heat sinks are a good choice when physical space is limited. Relative to active, air-based setups, passive systems don’t need as big an area. Additionally, they generate very little noise, making them preferable to active fan-powered systems when noise sensitivity is an issue.
Despite these plusses, passive systems still have some major shortcomings. They struggle to remove ambient heat at effective rates in certain scenarios; especially those involving high-performance components and heavy workloads. Passive heat sinks also struggle in environments with limited airflow and high density, where they simply cannot remove excess heat quickly enough to justify their use.
Liquid Heat Sinks
Liquid-based systems direct the heat generated by computing components into coolant fluids. These coolants absorb the incoming heat, conducting the thermal energy through radiator networks where it is eliminated.
Most liquid heat sinks are either open- or closed-loop systems. Open-loop systems require assembly, but are customizable. Closed-loop systems, on the other hand, come preassembled. However, reconfiguring or retrofitting a closed-loop setup can be difficult, or even impossible.
Among these variants, immersion cooling has continued to gain traction as a particularly effective liquid heat sink subtype. With immersion cooling, servers are submerged into precision-engineered coolant fluids that do not conduct electricity and therefore pose no risk of damaging hardware. The coolant fluids drastically impede the rates at which heat can build up in submerged servers, helping maintain their high performance and extend their lifespans.
Pros and Cons
Liquid cooling is especially useful in data centers. Higher levels of computing efficiency, reduced power consumption, and vastly superior cooling capacities make it preferable to both active and passive systems.
Additional advantages include:
- Greater efficiency
- Design flexibility
- A superior temperature impact and control profile
- Noise reduction
- Energy sustainability
Notably, liquid and immersion cooling systems are very useful in high-density data centers, which look set to become the norm thanks to the mainstreaming of data-intensive technologies like artificial intelligence, machine learning, and the blockchain.
That said, liquid immersion’s many strengths are tempered by a few issues that data center operators may wish to consider. For example, if the servers being used in the deployment have not been designed for immersion, there will be some conversion and modifications needed, likely including the use of new heat sinks designed for use in immersion environments. Also, some additional training and servicing and maintenance may be needed.
However, there are additional measures and configuration options that address these negative factors. In fact, once operational, immersion cooling heat sinks require very little electricity input. Their lower long-term operating costs hold the potential to generate positive returns on any necessary upfront investment.
Furthermore, artificial intelligence and machine learning technologies are increasingly being integrated into data center monitoring and maintenance systems. AI automates and streamlines many essential monitoring tasks related to heat sink conditions, making them easier to manage.
GRC Helps Data Center Operators Realize the Benefits of Liquid-Based Cooling Technologies
Active air-cooled heat sinks and passive systems still have their uses. Even so, their limitations are becoming more significant as data centers continue evolving toward greater density and complexity. Liquid-based systems and immersion cooling are rapidly emerging as favored heat management technologies in modern data centers, thanks to their superior capacity and performance characteristics.
Green Revolution Cooling (GRC) is a leading innovator and provider of liquid heat sinks and immersion cooling systems for data centers and other computing applications. To learn more, or to find bespoke solutions to your cooling needs, contact GRC.