Data centres as essential modern technology

Data centres are the backbone of the digital age, powering the vast networks of information that drive our interconnected world.

As workloads increase, they require more computing resources, and in turn, the demand for data centres increases. Last year’s Data Centre Trends 2023 suggested that demand for data centres will continue to grow by 15% year-on-year.

But it’s not all plain sailing. Data centres are, understandably, electricity and energy-intensive. There are major, widespread concerns as to their sustainability. As individuals, corporations, and countries the world over explore options towards net zero, data centres have frequently come under scrutiny.

For data centres to increase their energy efficiency, they need regulated, optimised computing performance, without needless expenditure and waste.

How do we guarantee that? And how do we make it effective?

The crux of the matter is cooling. One article suggests that ‘the average modern data centre uses around 40% of its power for cooling. But optimal conditions come at a cost. And that’s where sensors can truly come into their own.

Firstly — what are ‘hyperscale’ data centres?

Hyperscale data centres are big business. They’re highly scalable, high-performance environments typically built to support the requirements of cloud computing and big data processing. The power of Google, Amazon, Facebook, etc. to store, process, and deliver huge amounts of data (in real-time, no less) hinges on the robustness of their hyperscale data centres.

In short: these are quintessential 21st-century spaces, and we benefit from them daily.

What really sets them apart from conventional data centres is their high-density structure. Often, to maximise computing power and minimise space, hyperscale data centres utilise high-density computing equipment such as blade servers, storage arrays, and networking switches.

But within grand-scale compact tech, hotspots, dysfunction, and poor energy efficiency are all issues that intensify.

ARM solutions for hyperscale data centres

It’s certainly worth noting that several developers of hyperscale data centres are opting to build their own custom ARM clusters.

These CPU architectures have taken hyperscale computing to the next level, allowing centre owners and operators to load hundreds of CPUs on a single piece of silicon, whereas historically the industry has leveraged one CPU core per piece of silicon.

Such ARM solutions are making hyperscale data centres even more tightly packed and optimised for specific workloads, bringing greater heat density that renders miniature sensors such as Flusso’s all-the-more advantageous.

Hyperscale data centre cooling technologies

Temperature control and airflow management within server racks are the core methods for cooling hyperscale data centres.

The state-of-play: air cooling

Air cooling forms the basis of many data centres’ cooling technologies and strategies. Whilst emerging options such as liquid, immersion, and evaporation cooling are increasingly popular, air cooling remains the core industry focus, with innovations coming to the proven, reliable technology.

Air cooling uses air conditioning, fans, and vents to circulate ambient air, expelling the hot air produced by computing equipment. As the air cooling space evolves, temperature sensors remain a pivotal technology and airflow sensors emerge as a major asset in optimising temperature regulation.

How is air cooling deployed in hyperscale data centres?

Here are some core methods deployed by big businesses and site managers to effectively regulate hyperscale data centre temperature and energy efficiency.

• Computer Room Air Conditioning (CRAC): Large air conditioning units that draw in warm air, cool it using refrigeration systems, and then distribute it throughout the entire data centre space. These systems often draw on fine-tuned temperature sensor technology, and there is also scope for airflow sensors to be integrated. Incorporating airflow sensors can bring added value by increasing efficiency in cooling and air distribution for optimal performance.
• Alternating hot and cold air: Alternating racks with cold intakes facing one way and hot output facing the other helps hyperscale data centres to self-regulate and improve cooling efficiency. This approach is often combined with temperature sensors to maintain equilibrium across the alternate racks. Again, we are excited to be exploring how airflow sensors can optimise airflow distribution in the critical areas.
• Airflow management: Efficient airflow within server racks is crucial for maintaining optimal operating temperatures and preventing hardware failures. Flusso’s sensors are paramount to maintaining this airflow, and this is the principal use case in which we anticipate adding major value to the space.
• Raised floors: These create an underfloor plenum for air distribution, allowing cool air to flow up into the server racks and hot air to escape via ceiling vents. As with the previous technologies and techniques, flow sensors have the potential to come into their own by identifying current and required flow and feeding data to the ventilation system.

Flusso’s sensors are perfectly tailored to businesses that have yet to utilise airflow sensors in any respect, bringing standout, miniature sensor technology that can be mounted directly onto electronic boards.

Future trends and technologies

Data centre cooling technologies are fast-moving, with a big lean towards environmentally-conscious technology that will take us into the next few decades. Below, we’ve covered a select number of developments and explored how airflow sensors can be effectively deployed within them.

Liquid cooling

This technology utilises ultra-conductive liquids to quickly move heat away from servers and maintain temperature balance across hyperscale data centres. It is particularly effective for high-density facilities, with liquid cooling often offering superior heat removal as against air cooling, as well as a greater ability to effectively target hotspots within server racks.

Temperature sensors are vital here as they monitor thermal dynamics within the liquid cooling mechanism, optimising cooling by adjusting system components.

Immersion cooling

Immersion cooling is an increasingly popular data centre cooling technology that identifies hotspots through the full submersion of servers in dielectric fluid. The fluid cools the servers by absorbing heat rather than conducting it — reducing the reliance on other cooling infrastructures like fans and thus significantly reducing the potential energy consumption.

Direct-to-chip cooling

This data centre cooling technology is pretty simple: cool the server chips directly instead of cooling the entire server. This targeted approach enhances efficiency by reducing wasted energy expenditure sweeping servers with cooling efforts that aren’t required. Within this methodology, airflow sensors play a crucial role in monitoring and optimising airflow patterns within the servers.

Flusso’s precise, robust sensors will feed real-time data to the cooling system, ensuring direct-to-chip alterations are made exactly where they’re needed, thereby preventing the development of hotspots and improving overall system performance. The long-term benefit of sensors in such a technology is improved hardware lifespan and reduced maintenance costs in hyperscale data centre environments.

AI and machine learning for adaptive cooling

This serves as a sort of culmination to the several points above. Essentially, data centre cooling technologies are advancing at a rapid rate, and the systems at play are becoming more intelligent every year.

You can’t go far in business nowadays without mention of artificial intelligence (AI) and machine learning (ML). AI and ML algorithms have extraordinary potential in data centre management, including in the optimisation of cooling efficiency.

Vitally, utilising this ever-advancing technology also supports hyperscale data centres to further scale but with a keen eye on sustainable operations.

Renewable energy

We’re already seeing data centres that draw from renewable energy sources such as solar or wind power. Though reliably powering a hyperscale data centre is a little while off, renewable energy sources are an idyllic vision for the future of their development and maintenance. Such energy sources have the potential to offset energy consumption and facilitate innovative cooling technologies such as evaporative cooling or geothermal systems, which harness natural processes to maintain optimal temperatures without excessive energy consumption.

The incredible potential of airflow sensors

Whilst temperature sensors are currently the most commonly deployed sensors in several of the aforementioned data centre cooling techniques, airflow sensors have the potential to be more crucial over the coming decades.

Here’s the crux: deploying airflow sensors throughout the data centre infrastructure will allow for improved monitoring of airflow velocity, direction, patterns, and temperature gradients in real time, helping systems to cool and regulate the data centre’s environmental conditions with greater efficiency and accuracy.

Airflow sensors can be critical in revealing patterns and anomalies, hence triggering an alert that notifies operators of potential hotspots, equipment failures or blocked air vents before they escalate into critical failures. This proactive approach to maintenance can help minimise downtime and prevent costly disruptions to hyperscale data centre operations.

Detecting hotspots and optimising workload efficiency

Airflow sensors detect air circulation to reveal areas where heat accumulates due to uneven server loads or inefficient cooling. Flusso’s airflow sensors provide data that allows the ventilation system to dynamically redistribute workloads or adjust cooling resources to mitigate hotspots and mitigate the risk of a major outage as well as prolonging the lifetime of expensive systems.

The data that airflow sensors provide can be equally vital for predictive analytics, enabling proactive maintenance and capacity planning, thus reducing energy consumption.

Optimising energy costs

Everything we’ve covered thus far paints a clear picture: airflow sensors feed continuous, rich data to hyperscale data centre managers, thus allowing them to optimise spaces and maintain high-quality operating conditions. Subsequent dynamic adjustments reduce unnecessary cooling and associated power usage, hence reducing a business’s energy costs and carbon footprint.

Precise airflow and temperature data helps with capacity planning and lowers operational expenses whilst supporting core sustainability goals.

Conclusion

The evolution of hyperscale data centre cooling is underpinned by a pursuit of improved scalability, efficiency, and sustainability. There are several emerging trends and technologies that can bolster performance, from liquid and immersion cooling to AI-driven optimisation and renewable energy integration.

There are also several cases in which Flusso’s airflow sensors have the potential to address major challenges within densely packed hyperscale data centre spaces and enhance their efficiency going forward.