A. S. M. Hasan Mahmud
Florida International University School of Computing and Information Sciences
A. S. M. Hasan Mahmud is a Ph.D. candidate of the School of Computing and Information Sciences at FIU. He received his Bachelor’s degree in Computer Science and Engineering in December 2007 from Bangladesh University of Engineering and Technology, Bangladesh. He obtained his Master degree in Computer Science from FIUÛªs School of Computing and Information Sciences in 2014. Hasan is currently supervised by Dr. Shaolei Ren and co-advised by Dr. Iyengar. His research interests include cloud computing and data center resource management, with an emphasis on sustainability. He received the best paper award from the 8th International Workshop on Feedback Computing (in conjunction with USENIX ICAC, 2013).
To meet the surging demand for online computing, data centers are continuously growing in both numbers and sizes, increasing their electricity consumption and carbon footprints worldwide. Today’s large data center hosts hundreds of thousands of servers and its peak power rating exceeds 100MW. Currently, the data centers consume 3% of global electricity production and would rank 5th in the world if the data centers were a country. A significant portion of this electricity is produced form carbon-intensive sources (e.g., coal and oil), often called “brown energy”. Due to the brown energy consumption, data centers are accountable for emitting 200 million metric tons of carbon dioxide per year. Many IT organizations (e.g., Apple, Facebook, and Google) are consistently pressured, both from utility providers and governments, to reduce their carbon footprint and energy consumption. While these companies have taken steps to reduce their carbon footprints (e.g., by installing on-site/off-site renewable energy facility), they are consistently looking for new approaches to reduce their energy consumption and carbon footprints without incurring significant operational costs.
The numerous existing studies on reducing the carbon footprint and energy consumption of a data center mainly focus on self-managed data centers. While these studies are encouraging and proven effective, they cannot be applied to a hybrid data center infrastructure because of its major architectural and operational differences. Our study takes the first step to incorporate the diverse cost and energy management schemes of a hybrid data center infrastructure. This research aims to identify the key challenges and explore the resource management for a hybrid data center infrastructure from two aspects: (1) how to reduce the carbon emission of a hybrid data center infrastructure while satisfying the performance requirements and without incurring significant additional operational cost, and (2) how to reduce the energy consumption of a hybrid data center infrastructure via temperature aware workload scheduling. We propose to develop algorithms to solve the aforementioned problems.