Distributionally Robust Games for Data Center Demand Response Coordination based on CPU utilization
Cloud providers seek effective solutions to reduce power consumption and ease grid stress for data centers (DCs). Dynamic voltage and frequency scaling (DVFS) could reduce CPU power consumption and provide demand response. However, its real-time performance depends on uncertain and volatile CPU utilization rates, which highly undermine demand response service quality. We present a scalable distributionally robust game framework for the aggregator to coordinate DC for DVFS demand responses. Using millions of CPU readings from Microsoft virtual machines, games are performed with varying numbers of players. When game solutions exist, both models achieve the required demand reduction while considering the quality of service (QoS) for each data center (DC). The equivalent optimization model could allow a 100-player game to be solved on a laptop in under a minute.
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
- 323 Dr Martin Luther King Jr Blvd
- Newark, New Jersey
- United States 07102
- Building: ECE 2nd floor
- Room Number: 202
- Contact Event Hosts
- Co-sponsored by Power Systems Engineering Center (PSEC) at New Jersey Institute of Technology
- Starts 14 July 2025 05:00 AM UTC
- Ends 21 July 2025 03:00 PM UTC
- No Admission Charge
Speakers
Dr. Yifu Ding of MIT Energy Initiative
Distributionally Robust Games for Data Center Demand Response Coordination based on CPU utilization
Cloud providers seek effective solutions to reduce power consumption and ease grid stress for data centers (DCs). Dynamic voltage and frequency scaling (DVFS) could reduce CPU power consumption and provide demand response. However, its real-time performance depends on uncertain and volatile CPU utilization rates, which highly undermine demand response service quality. We present a scalable distributionally robust game framework for the aggregator to coordinate DC for DVFS demand responses. Using millions of CPU readings from Microsoft virtual machines, games are performed with varying numbers of players. When game solutions exist, both models achieve the required demand reduction while considering the quality of service (QoS) for each data center (DC). The equivalent optimization model could allow a 100-player game to be solved on a laptop in under a minute.
Biography:
Yifu joined the MIT Energy Initiative as a postdoctoral research associate in January 2023. She is currently a senior research associate with the MIT Energy Initiative and the MIT Sloan School of Management. She holds a Ph.D. in Engineering Science from the University of Oxford. Her research centers on machine learning forecasting and advanced optimization techniques that account for uncertainty in power system operations and planning. Yifu has received several honors, including the 2024 MIT Open Data Prize and the 2025 MIT Energy Scholar Fellowship.
Email: