Data centre optimisation essential in the age of AI

Joleen Maritz, IT Enclosure Sales, Rittal South Africa. (Image: Supplied)

---

Artificial intelligence (AI) is transforming the business landscape, optimising efficiency and driving productivity. By automating tasks and enabling data analytics, it can provide businesses with a significant competitive edge. However, as Joleen Maritz, IT Enclosure Sales, Rittal South Africa explains, the rapid adoption of AI is prompting significant advances in data centre infrastructure. “To successfully deploy AI tools, we need data centres with the capacity to train advanced AI models,” Maritz says.

Maritz points to the findings of The Data Centres in Africa report, which suggests that the data centre market will increase by 50% by 2026. “Data centre energy consumption is also expected to grow rapidly, driven by the adoption of AI technology. Training of AI models requires large quantities of data and processing power.”

For data centre operators, managing the rise in thermal loading within racks will become a priority. “Electrical components are getting smaller, but rack densities are increasing, causing more heat to be trapped inside racks. There’s a real risk of damage to the sensitive electronic components,” notes Maritz.

For data centres with racks that have a lower power requirement, air cooling (in-aisle) may be sufficient, but Maritz reiterates that liquid cooling delivers the greatest benefits.

Liquid cooling packages are proving effective in high-density facilities (where rack loads are reaching in excess of 40kW-50kW) by delivering cooling directly where it’s most needed. “Rittal’s liquid cooling solutions are energy efficient and scalable, easily expanding to accommodate data centre expansion,” adds Maritz.

Maritz cautions that in racks where heat densities exceed 200kW, air cooling is not recommended. “Immersion cooling or on-chip cooling are more effective options.”

Immersion cooling involves the submerging of servers into tanks containing a dielectric fluid. While this solution is efficient, it is complicated, introducing the need for more robust structural floors to support the additional weight of the tanks (a possible threat to equipment warranties). “On-chip cooling, on the other hand, is where liquid coolant is pumped through pipes directly to heatsinks on the chips, to absorb excess heat. This method reduces the data centre’s energy consumption and has been shown to improve system uptime,” says Maritz.

A further concern around the rapid adoption of AI is the demand for optimal data processing. “Companies are increasingly integrating AI models into their operations, prompting demand for fast and accurate real-time data processing. To ensure this is achieved with minimal latency, IT equipment needs to be placed close to the source of data generation. But placing delicate electronic components in harsh operating environments – such as a dusty factory floor – can be risky,” explains Maritz. “This is driving more businesses to deploy edge data centres, which are typically quick to install and come with built-in cooling, power supply and IT racks.”

Rittal offers an extensive range of enclosure and IT rack solutions for edge deployments. Featuring modular designs that makes installation and expansion easy, these are robust and reliable. “The expansion of our footprint, both locally through the establishment of regional branches, as well as throughout Africa, enables us to extend our support, expertise and product offering to our customers, anywhere,” says Maritz.

“We can’t underestimate the potential of AI technology to drive business advances. Yet to enable the adoption of AI, data centre operators need to invest in infrastructure that will support the energy and cooling demands of their modern data centres,” concludes Maritz. “Rittal has invested in the development of solutions to future-proof IT systems and help businesses optimise the performance and longevity of their IT equipment, ensuring our customers can make the most of new technologies.”