George Papastefanatos, Principal Researcher at the Information Management Systems Institute of the Athena Research Center, gave a highly insightful interview on the "Review" program, broadcast on Naftemporiki’s channel, with journalist Takis Spiliopoulos. The discussion focused on data center sustainability and innovative methods to reduce energy consumption.
In recent years, the increasing use of technology has significantly raised computing resource demands, making the search for more efficient and "green" solutions essential. Within this context, Mr. Papastefanatos and his research team have developed machine-learning-based methods aimed at reducing energy consumption in data centers through intelligent management of available computing resources.
The ARCADIA research project, implemented by Mr. Papastefanatos’ team, takes into account the fluctuations in demand for modern applications used daily. These fluctuations necessitate the dynamic adaptation of computing resource usage. "We needed to develop algorithms capable of predicting, throughout the day, when demand would be higher or lower, allowing for system scaling up or down" explains Mr. George Papastefanatos.
The second part of their research focused on managing available machines so that they are allocated or deallocated based on demand forecasts. Through reinforcement learning techniques, the algorithms optimize which machines remain active and which stay idle, ultimately reducing energy consumption.
Estimates indicate that implementing these technologies could lead to significant energy savings, substantially contributing to lowering the energy footprint of data centers.
The energy consumption of data centers is a growing concern as technology becomes increasingly embedded in our daily lives. The ARCADIA research team strives to make computing resource management smarter, reducing energy waste and optimizing system performance.
The transition to greener data centers is not just a technological challenge but a necessity for a sustainable digital future.