Flick the AI switch, turn up data centre sustainability

One of the main purposes of technology is to make our lives more efficient, from generative AI powering human-like conversations right through to Internet of Things (IoT) sensors and robotics managing warehouse distribution. These examples are small glimpses into the technologies that have been designed to complement and simplify our lives.

But behind the scenes, what it takes to power these technologies is pushing up energy spend and ESG pressure.

Data centres are the essence of our burgeoning digitised world, and therefore consuming more and more energy. This has put data centres in the firing line for the growing percentage of grid energy they use.

But according to the Australian Energy Board, the average Australian data centre is 20 years old and often inefficiently designed. While colocation facilities typically rely on newer equipment, the two-decade lifetime is certainly playing out in enterprise today and creating an energy and environmental burden.

This presents challenges not only for operators on the ground, but also in the boardroom where ESG is a top priority.

Lowering data centre energy consumption requires identifying and managing the cost and performance of associated investments, as ultimately, they must make sense to a very wide range of stakeholders throughout an organisation.

Sustainability is everybody’s business

Long-held conversations around efficiency and utilisation in the data centre must evolve to reflect a more comprehensive and aggressive focus on sustainability.

In a Schroders global sustainability investment report, there are three factors halting boardroom agreement on sustainable investment. First, there is often a lack of transparency and reported data about the impact of investments. Second, there is absence of an agreed definition on what sustainable investment is and third, there are performance management concerns.

At the decision-makers’ table, there are emotional drivers aplenty for decarbonisation, but organisational efforts will only succeed for a good business reason. And that reason must meet the requirements and motivations of the many members of the C-suite.

While the reality is more nuanced, put plainly, the CIO is driven to keep their technology ecosystem lean without compromising resilience and reliability; meanwhile, the COO expects investments to improve availability and service for better customer experience. CFOs want to see cost efficiency and ROI, while CEOs sit across it all and are mindful of lip service initiatives and potential brand harm.

The Australian Securities and Investments Commission (ASIC) named “action against greenwashing” one of its priorities for 2023. Window dressings will only see ASIC come knocking, and that’s not a smart business decision nor a healthy one for the planet.

The entire C-Suite has more reason than ever to apply green strategies. Getting it right means lowering carbon output, reducing energy spend without compromising availability and resilience, and providing tangible datasets to monitor ongoing real impact on carbon emissions.

Machine learning to tune energy optimisation

Our data-hungry nature isn’t going to change. Reliance on data centres is only increasing and the numbers are representative — Australia is expected to double to 417,000 m² over the next three years with our country the fastest growing data centre market in Asia–Pacific.

So, why not apply one of the key and fastest-growing technologies data centres power — AI and machine learning (ML) — to data centres themselves?

One Australian startup has developed an innovation that prompts ML systems to apply automatic modulation to chiller units — it’s working in over 100 data centres and has achieved energy savings of between 8 and 16%.

With cooling systems, which are the biggest overhead target for efficiency, intelligently adapting output in response to the learnt environmental factors for ongoing real-time optimisation is game changing. And it typically pays for itself in under three years.

The system also has performance management built in, meaning from design to delivery it tracks carbon footprint, managed together from a cost and efficiency perspective. It’s like a pair of eyes that keep a keen watch on every aspect of the facility, providing an unparalleled level of insight into energy usage and efficiency, and overall ROI.

Turning back to Schroder’s three barriers to sustainable investment — accountability, definition and performance — this one example proves AI and ML in the data centre tick all boxes, while speaking to the primarily C-level motivations.

The benefit of this technology is its proven ability to make fine-tuned machine learning-based adjustments to the primary cooling systems, ensuring they are continuously and efficiently responding to data centre demands and returning with the most energy-efficient response to service level demands.

Ultimately, it’s up to us to embrace these innovative technologies and take an active role in reducing energy consumption and contribute to a more sustainable future. After all, data centres store all our brightest ideas.

*Peter Simon is IT Solutions Architect Asia Pacific and Robert Linsdell is Managing Director Australia and New Zealand at Vertiv.

Image credit: iStock.com/Just_Super