Enhancing Energy Efficiency in Kubernetes Cluster Through Resource and Energy Aware Scheduling

As the demand for cloud-native infrastructure grows, managing energy becomes increasingly essential for small-scale and large-scale Kubernetes clusters, including those within the IoTcloudServe@TEIN environment. The research introduces a resource and energy aware scheduler to optimize energy usage by dynamically managing node status and workload assignment. By considering workload demands, the scheduler enables worker nodes to be activated and deactivated, optimizing energy efficiency and resource allocation. The proposed algorithm is implemented as an extender on top of the default Kubernetes scheduler. The study shows the performance of the proposed scheduler against the default Kubernetes scheduler and the bin packing scheduler by focusing on energy and power utilization metrics. Experimental results reveal that the proposed resource and energy aware scheduler can significantly reduce energy consumption by up to 17% compared to the default Kubernetes scheduler and up to 18% compared to the bin-packing scheduler.