Towards Carbon-Aware Container Orchestration: Predicting Workload Energy Consumption with Federated Learning

TL;DR

The paper tackles the privacy and generalization issues of centralized energy-prediction models for carbon-aware container orchestration. It presents a federated learning framework that extends Kubernetes Kepler to train XGBoost models across distributed clients using FedXgbBagging, preserving data locality. On the SPECPower benchmark, the federated approach achieves an 11.7% reduction in MAE compared with a centralized baseline, demonstrating strong predictive accuracy while maintaining privacy. This work highlights federated learning as a viable path for sustainable cloud computing, enabling carbon-aware scheduling without centralized data sharing and with potential for real-world deployments in heterogeneous data-center environments.

Abstract

The growing reliance on large-scale data centers to run resource-intensive workloads has significantly increased the global carbon footprint, underscoring the need for sustainable computing solutions. While container orchestration platforms like Kubernetes help optimize workload scheduling to reduce carbon emissions, existing methods often depend on centralized machine learning models that raise privacy concerns and struggle to generalize across diverse environments. In this paper, we propose a federated learning approach for energy consumption prediction that preserves data privacy by keeping sensitive operational data within individual enterprises. By extending the Kubernetes Efficient Power Level Exporter (Kepler), our framework trains XGBoost models collaboratively across distributed clients using Flower's FedXgbBagging aggregation using a bagging strategy, eliminating the need for centralized data sharing. Experimental results on the SPECPower benchmark dataset show that our FL-based approach achieves 11.7 percent lower Mean Absolute Error compared to a centralized baseline. This work addresses the unresolved trade-off between data privacy and energy prediction efficiency in prior systems such as Kepler and CASPER and offers enterprises a viable pathway toward sustainable cloud computing without compromising operational privacy.

Figures (4)

• Figure 1: Federated learning architecture
• Figure 2: Federated XGBoost Bagging Aggregation Strategy
• Figure 3: Mean Absolute Error Plot comparing the baseline centralized XgBoost vs the federated XgBoost Model
• Figure 4: Evaluation metrics for federated XGBoost model (averaged over 3 independent runs).