EcoKube simulates carbon-aware scheduling for edge-cloud sustainability
New framework lets you test sustainability policies before deployment—no real infrastructure needed.
EcoKube tackles the growing energy demand from cloud and edge computing—especially AI workloads—by enabling reproducible evaluation of carbon-aware scheduling policies. The framework simulates time- and location-varying grid Carbon Intensity (CI), site Power Usage Effectiveness (PUE), and heterogeneous hardware characteristics across federated edge-cloud topologies. It provides an event-driven deterministic simulator, policy hooks for custom schedulers, and a heterogeneity-aware reference policy. Researchers can test spatio-temporal workload shifting, temporal elasticity, and carbon-aware placement without needing real distributed infrastructure.
In their evaluation, the authors compare EcoKube's reference policy against three baselines: the default Kubernetes scheduler, KEIDS, and TOPSIS/KCSS. Using synthetic batch workloads, they demonstrate how the framework can quantify sustainability trade-offs. The paper (6 pages, 2 figures) was submitted to the TDIS 2026 workshop at EuroSys2026. EcoKube's key architectural contribution is enabling consistent, reproducible comparisons of sustainability-aware scheduling policies before deployment, helping data centers and edge providers optimize for lower carbon footprints.
- EcoKube uses an event-driven deterministic simulator with pluggable policy hooks for testing carbon-aware scheduling
- Framework models grid Carbon Intensity (CI), Power Usage Effectiveness (PUE), and heterogeneous hardware across sites
- Evaluation compares against Kubernetes scheduler, KEIDS, and TOPSIS/KCSS using synthetic batch workloads
Why It Matters
Gives cloud and edge operators a reproducible way to cut carbon emissions before deploying new scheduling policies.