Researchers from Nanyang Technological University, Singapore (NTU Singapore) and its spin-off enterprise Red Dot Analytics (RDA) have worked together to develop innovative technologies that cut data centre energy consumption and emissions. The modern digital economy is dependent on data centres, which are currently confronted with several issues, including growing energy costs, stiffer rules on carbon emissions, and the rapid expansion of cloud computing.
Like the equivalent of the metaverse, I envision a Data Centre metaVerse (DCverse), where all data centres are controlled virtually by AI agents.
– Professor Wen Yonggang, Associate Dean-Research, College of Engineering
Professor Wen shared that human operators can remotely control the physical hardware of a data centre in cyberspace, with robots acting as their proxies to implement these modifications. This might create countless opportunities for data centres, such as putting them in cooler environments, such as underground or underwater, where less energy is required for cooling.
Using Artificial Intelligence and digital twins – a full-scale replica of a data centre’s physics and operations in the virtual world – RDA can assist businesses in evaluating the entire life cycle of their operations to ensure a holistic assessment of carbon emissions and energy consumption, and to validate new changes and processes prior to implementing them in the real world.
RDA can optimise the operations of a data centre, increase its stability and performance, and reduce energy consumption by up to 30%, resulting in a substantial decrease in electricity bills and associated carbon emissions.
The country’s efficient digital and electricity infrastructure, few natural risks, business-friendly environment, and trained workforce make it a desirable location for data centres, however, the warm tropical climate provides issues for data centre operations, especially with global warming that resulted in half of tropical data centre energy use going to cooling.
A data centre’s architecture and operations will be examined from beginning to end using RDA’s revolutionary Performance and Sustainability Lifecycle methodology. RDA may then increase operation efficiency using approaches like predictive maintenance, capacity planning, and dynamic workload and cooling allocation based on the vast amount of data gathered.
For example, if server loads can be dispersed evenly in real-time via AI optimisation, the servers will generate less heat and energy will be saved by avoiding the common practice of over-cooling, which is used to accommodate unexpected peak demands.
An overall reduction in data centre electricity usage will help Singapore achieve its climate goals of net zero emissions by mid-century, as revealed in March’s Budget 2022.
Recently, Singapore has also lifted a prohibition on new data centres that had been in effect since 2019 and established a pilot phase of new sustainable data centres that must contain innovations and sustainability solutions. This is in line with the government’s goal of establishing data centres that are best-in-class in terms of resource efficiency and contribution to economic and strategic goals.
Meanwhile, research at NTU Singapore is still ongoing to develop a “digital-first” approach for data centre designs. Instead of designing a prototype data centre and building it in the real world to verify its performance, Prof. Wen and his team created a framework and software for a virtual data centre. AI agents optimise the architecture and operations to achieve the finest virtual data centre that can be used to establish and control the real one.