Carbon-aware Multi-Site Workflow Scheduling Under Uncertainty (B01)
Abstract
In phase I of FONDA, we researched methods to schedule and execute DAWs across heterogeneous infrastructures for improving adaptability and portability. We developed profilers and micro-benchmarks that analyze the cluster infrastructure and expose those metrics to our map-ping and scheduling methods. Further, we designed a novel approach to provide runtime estimates without historical monitoring and new methods for mapping tasks to heterogeneous infra-structures. Yet, the global need to reduce carbon emissions calls for new and carbon-aware optimization when scheduling over multiple data centers. Thus, in phase II we research adaptive multi-center scheduling methods under uncertainty, focusing on carbon footprint minimization of scientific DAWs.
Principal investigators
Participating external organisations
Financer
DFG Collaborative Research Centre
Duration of project
Start date: 07/2024
End date: 06/2028
Subproject of
07/2024 - 06/2028 |
