Researchers from Red Dot Analytics (RDA), a spin-off company of Nanyang Technological University, Singapore (NTU Singapore), and NTU Singapore have collaborated to develop novel solutions that reduce emissions and energy use in data centers. Data centers, which are essential to the modern digital economy, are currently struggling with a number of problems, including rising energy costs, stricter regulations on carbon emissions, and the quickening rise of cloud computing.
I picture a diverse, data center counterpart of the metaverse, where all data centers are practically run by AI agents – Prof. Wen Yonggang, associate dean for research in the engineering college
Professor Wen explained that by using robots as their proxies, human operators can remotely modify the physical hardware of a data center while it is located in cyberspace. This might open up a wide range of possibilities for data centers, such as placing them in colder locations where cooling requires less energy, like underground or underwater.
RDA can help companies evaluate 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 before putting them into practice in the real world. These technologies include artificial intelligence (AI) and digital twins, which are full-scale replicas of a data center’s physics and operations in the virtual world. RDA can optimize a data center’s operations, improve stability and performance, and consume up to 30% less energy, which will significantly save electricity costs and related carbon emissions.
The country is a desirable location for data centers due to its effective digital and electricity infrastructure, low risk of natural disasters, business-friendly environment, and skilled workforce; however, the warm tropical climate presents challenges for data center operations, particularly with global warming that has caused half of tropical data centers’ energy use to go to cooling.
Using RDA’s ground-breaking Performance and Sustainability Lifecycle methodology, the architecture and operations of a data center will be reviewed from start to finish. Based on the huge amount of data collected, RDA may then optimize operational efficiency utilizing techniques like predictive maintenance, capacity planning, and dynamic workload and cooling allocation.
For instance, if server loads can be equally distributed in real-time by AI optimization, the servers would produce less heat and energy will be conserved by avoiding the common practice of over-cooling, which is utilized to meet unanticipated peak demands. As stated in March’s Budget 2022, Singapore’s climate goals of net zero emissions by the middle of the century can be met with a reduction in overall data center electricity consumption.
A ban on new data centers that had been in place since 2019 was recently lifted in Singapore, and a pilot program for innovative and sustainable data centers has also been launched. This supports the government’s objective of building data centers that are best-in-class in terms of resource efficiency and contribution to economic and strategic goals.
Research at NTU Singapore is still being done to create data center architectures that are “digital-first.” Prof. Wen and his team developed a framework and software for a virtual data center as opposed to establishing a prototype data center and constructing it in the real world to evaluate its performance. To create the best virtual data center that can be used to establish and control the real one, AI agents optimize the architecture and operations.
