30 October 2024
Nick Ewing, Managing Director, EfficiencyIT
There's no one-size-fits-all when it comes to data centre cooling. The optimal solution will always be application-specific and based on the customers’ requirements.Historically, air cooling has been the primary method used to maintain optimum cooling temperatures within data centres. However, the growing demand for AI and GPU-accelerated computing has increased rack densities and power consumption, and increased the heat generated by the servers, necessitating more effective cooling strategies such as liquid cooling.
Liquid cooling, particularly direct-to-chip and immersion-based solutions, are gaining popularity due to their superior performance in handling the high heat demands associated with these modern AI and HPC workloads. Hybrid cooling systems, combining air and liquid cooling, are also emerging as a preferred and flexible option, providing the necessary balance between cost, efficiency and performance in shared environments. Here, direct-to-chip liquid cooling is the perfect fit.
Key considerations when purchasing new equipment
When purchasing new cooling equipment, both the present and future needs of the enterprise should be evaluated.
It's essential to consider where your technology roadmap is heading. If your business expects to support high-powered GPU equipment or AI-based applications in the near future, it's paramount to invest in a cooling solution that can accommodate those high-performance demands.
In particular, investing in liquid cooling technologies can future-proof your data centre, ensuring you're not left scrambling to upgrade your cooling requirements as your IT infrastructure evolves.
Watch out for pieced-together systems
We've seen many enterprises opt for piecing together equipment from multiple manufacturers. While this might seem cost-effective initially, it often results in inefficiencies, lower resilience, and higher overall maintenance costs. A turnkey solution, built with a unified, best-in-class approach, will be more efficient, reliable, and sustainable in the long term.
It's essential to choose systems that meet your performance and energy efficiency goals. This will reduce operational costs and ensure your data centre remains competitive and environmentally optimised.
Latest innovations
The landscape of cooling technologies is constantly evolving. The latest innovations include advanced liquid cooling techniques like direct-to-chip and precision immersion cooling. These technologies ensure that critical components, such as CPUs and GPUs, are kept within optimal temperature ranges, maximising performance and minimising energy consumption.
Direct-to-chip cooling is especially promising, allowing enterprises to mix air—and liquid-cooled systems within the same data centre and seem to be preferred by colocation and hyperscale organisations.
Balancing costs with benefits
A frequent question we encounter is: How much should I spend on cooling technology? The truthful answer is it depends. Price points will vary based on the manufacturer, the solution, and the relationship your technology partner has with its vendors.
It's easy to overspend on features you may not need, but under-investing can be equally problematic, leading to higher long-term costs in energy usage, maintenance, and potential downtime.
Like most things in the industry, the key is to find the right balance. Ensure that a higher-cost solution's additional features genuinely add value to your specific application.
Warranties and guaranteesWhen investing in a new data centre cooling system, or any new IT or power system for that matter, securing long-term warranties and maintenance or service agreements is fundamentally crucial. We always recommend purchasing an extended warranty with any new technology. This strategic decision ensures you can quickly address unforeseen issues, reducing potential downtime and safeguards your organisation by minimising operational risks.
Opting for a maintenance service agreement can provide additional peace of mind. Your cooling systems will require regular checks and optimisations, and having a dedicated team on call will help avoid any unnecessary disruptions so that your data centre can continue powering the business-critical applications it’s hosting.
