Datacenter energy consumption represents about 1% of worldwide electricity consumption. With demand for datacenter computing projected to double every few years, this is simply not sustainable without continued rapid improvements
in datacenter energy efficiency.
Our vision in this proposal is to achieve a quantum leap in datacenter energy efficiency and resource utilization, by proposing Treehouse, a system architecture for balancing resource consumption and allowing developers to optimize
for lower energy usage. The key design principle underlying Treehouse is to treat workloads and hardware resources in a fine-grained and fungible way. Importantly, Treehouse also provides developers the means to precisely estimate the energy usage of their applications and implement scheduling and hardware offloading policies that optimize for lower energy usage.
Treehouse: A Case For Carbon-Aware Datacenter Software, Thomas Anderson, Adam Belay, Mosharaf Chowdhury, Asaf Cidon, Irene Zhang. Arxiv.
HotCarbon 2022: 1st Workshop on Sustainable Computer System Design and Implementation
Faculty / senior researchers: Tom Anderson, Adam Belay, Mosharaf Chowdhury, Asaf Cidon, Irene Zhang.
PhD students: Yuhong Zhong
MS students: Rahul Chaudhari, Rishabh Gupta
Fall 2021: class project in EECS 6897, distributed storage systems