GreenKube: Towards Greener Container Orchestration using Artificial Intelligence

This paper introduces the GreenKube framework, which aims to reduce energy consumption while meeting Quality of Service (QoS) requirements through the use of various AI methodologies such as deep learning time-series forecasting, deep reinforcement learning, and graph neural networks. The paper also explores the limitations of contemporary container orchestration frameworks, including Kubernetes, and describes how GreenKube aims to progress beyond them. Additionally, the paper presents a prototype of the GreenKube framework that was evaluated in an extensive simulation using CloudSim Plus, and compares its performance to Kubernetes' Horizontal Pod Autoscaler. The results demonstrate that GreenKube outperforms Kubernetes in terms of latency and task execution time while requiring fewer computational resources.