TY - GEN
T1 - Carbon-Aware Spatiotemporal Scheduling of Data Transfers
AU - Goldverg, Jacob
AU - Rodrigues, Elvis
AU - Kosar, Tevfik
N1 - Publisher Copyright:
© 2026 Copyright held by the owner/author(s).
PY - 2026/6/22
Y1 - 2026/6/22
N2 - The rapid growth of workloads such as AI, cloud computing, and big data analytics has made sustainable computing practices increasingly critical. Spatiotemporal workload shifting within and across data centers has proven to be a viable solution to reduce the carbon footprint of computing by enabling workloads to be delayed and migrated according to the availability of renewable energy. However, most existing work in this area overlooks the carbon cost of data transfers and network utilization, leaving a significant source of emissions unaccounted for. In this paper, we present CATS, a carbon-aware spatiotemporal data transfer scheduler, that accounts for end-to-end carbon emissions. We show that CATS can reduce carbon emissions by up to 79% compared to baselines by jointly shifting transfers across both time and network paths while maintaining high job completion rates under heavy network load.
AB - The rapid growth of workloads such as AI, cloud computing, and big data analytics has made sustainable computing practices increasingly critical. Spatiotemporal workload shifting within and across data centers has proven to be a viable solution to reduce the carbon footprint of computing by enabling workloads to be delayed and migrated according to the availability of renewable energy. However, most existing work in this area overlooks the carbon cost of data transfers and network utilization, leaving a significant source of emissions unaccounted for. In this paper, we present CATS, a carbon-aware spatiotemporal data transfer scheduler, that accounts for end-to-end carbon emissions. We show that CATS can reduce carbon emissions by up to 79% compared to baselines by jointly shifting transfers across both time and network paths while maintaining high job completion rates under heavy network load.
KW - carbon-aware scheduling
KW - Data transfers
KW - mixed-integer linear programming
KW - spatiotemporal scheduling
KW - workload shifting
UR - https://www.scopus.com/pages/publications/105044943779
U2 - 10.1145/3797248.3815406
DO - 10.1145/3797248.3815406
M3 - Conference contribution
AN - SCOPUS:105044943779
T3 - Proceedings of the 16th ACM International Green and Sustainable Computing Conference, IGSC 2026
SP - 36
EP - 41
BT - Proceedings of the 16th ACM International Green and Sustainable Computing Conference, IGSC 2026
A2 - Chen, Fan
A2 - Zhou, Peipei
A2 - Zand, Ramtin
A2 - Roohi, Arman
A2 - Hu, Jingtong
A2 - Yang, Xiaoxuan
PB - Association for Computing Machinery, Inc
T2 - 16th International Green and Sustainable Computing Conference, IGSC 2026
Y2 - 22 June 2026 through 24 June 2026
ER -