Balancing high-performance computing with sustainability

The world around us has become increasingly connected — from working from anywhere to educating children, to delivering rapid patient diagnoses. Essential to our digital world, therefore, are the data centres that hold vast amounts of information being generated around the clock. Virtually every industry, be it banking, logistics, manufacturing or shipping, is investing in technologies to help extract value from data. The winners are those who can do it faster, more efficiently and more securely than the competition.

Organisations require enormous processing capacity to stay competitive, many turning to high performance computing (HPC) and artificial intelligence (AI) technologies, which are being deployed on-premises, in the cloud and at the edge.

Experts predict that the global HPC market is expected to reach AU$92 billion (US$60 billion) by 2025. Researchers have also estimated that the market size value of AI is expected to reach AU$218 billion (US$136.6 billion) by the end of 2022.

We are witnessing specific sectors gaining significant benefits from utilising HPC and AI technologies, such as manufacturing, which is using HPC and AI for predictive and prescriptive maintenance, automation of product lifestyle management and short design cycles. HPC and AI technologies are also being used in the healthcare industry to accelerate the development of new drug discovery, and in meteorology for more precise weather forecasting and improved understanding of climate change.

At the same time, organisations are under pressure to achieve net zero emissions by the year 2050 at the latest. Data centre operators must prioritise energy efficiency, especially as the demand for HPC and data storage grows.

With this in mind, what solutions can we use in data centres to improve sustainability, while still maintaining customer service and business profitability?

Reducing operational carbon

Data centres represent up to 4% of Australia’s total energy consumption and approximately 10% globally, according to Aurecon. That’s why data centre operators are working hard to drive down energy consumption to advance sustainability in their facilities.

Faced with increasing energy prices and rising pressure from external stakeholders to reduce carbon footprints, setting sustainable energy goals is a fundamental step in the data centre design process.

Enter HPC cooling solutions. They are more energy and resource efficient than a traditional approach because they include airflow management or use water as a cooling medium and can be in direct contact with the heat source. This is known as direct liquid cooling (DLC). One benefit of liquid cooling is the increased performance of the overall system due to lower latency of higher-density systems. For example, DLC results in total energy savings of 10% from eliminating the servers’ fans used in traditional air cooling.

HPC infrastructure also copes with higher water or air temperatures than traditional systems, which helps to cool the equipment and reduce electricity consumption. This helps reduce the data centre’s total carbon footprint. Additionally, the return water temperature from liquid-cooled racks is hot enough to supplement some building heating systems, which can offset a portion of the energy that buildings consume to heat offices and kitchens.

Clearly, HPC offers a way to improve energy efficiency and drive optimal performance in the data centre. Look no further than HPE’s HPC solutions, which ranked 1st, 2nd, 3rd and 6th on the 2022 Green500 list.

Pursuing additional energy and cost savings

Putting a green lens on the operational aspects of data centre sustainability is only the first step.

Organisations also need to focus on embodied carbon, which is the total greenhouse gas emissions resulting from everything in the upstream supply chain that was required to construct the data centre facility. This includes mining, manufacturing, transportation and installation of building materials.

For example, data centres use large amounts of concrete and steel, whose manufacturing processes are responsible for large quantities of carbon dioxide emissions. Addressing embodied carbon in the construction phase will help achieve carbon neutrality.

Because HPC can be built more densely, it reduces space requirements and thus requires fewer building materials. Fewer building materials means less embodied carbon. Similarly, smaller data centres require less cooling infrastructure, which also helps reduce embodied carbon emissions associated with building a new data centre.

Aside from the environmental savings, the reduced building materials also reduces capital costs. For example, decreasing the raised floor area from 930 square metres to 185 square metres will cut the cost of the facility by up to 20%. To put this in perspective, a typical cost for a facility of this size with power and cooling systems would be a minimum of AU$30.7 million (US$20 million), so a 20% reduction would equate to AU$6 million (US$4 million).

Looking ahead

HPC and AI technology are paving the way for sustainable data centre design that enables the least amount of infrastructure for power, cooling and space. This reduces both operational and embodied carbon while driving optimal performance. Now is the time for organisations to embrace HPC and AI technology as a step towards achieving sustainability goals.

Image credit: iStock.com/undefined