MACHINE LEARNING-BASED ENERGY OPTIMIZATION FOR CONTAINERIZED CLOUD WORKLOADS

Abstract

Containerized cloud workloads have become the foundation of modern enterprise cloudnative applications because they provide scalability, portability, fault tolerance, and efficient resource utilization. However, increasing deployment density and dynamic workload behavior have significantly increased data center energy consumption, making energy optimization an important challenge for cloud service providers. Conventional resource allocation approaches primarily optimize computational performance and application availability without explicitly considering energy efficiency or environmental sustainability. This paper proposes a machine learning-based energy optimization framework for containerized cloud workloads by integrating Kubernetes orchestration, predictive analytics, cloud-native monitoring, machine learning, and green computing. The proposed framework continuously analyzes workload characteristics, infrastructure utilization, application demand, energy consumption, and Quality of Service (QoS) metrics to generate intelligent resource optimization strategies. Experimental analysis demonstrates significant improvements in energy efficiency, infrastructure utilization, workload balancing, operational cost reduction, and sustainable cloud operations compared with conventional cloud resource management approaches. Keywords: Machine Learning, Energy Optimization, Containerized Workloads, Kubernetes, Cloud-Native Computing, Predictive Analytics, Green Computing, Resource Optimization, Sustainable Cloud, Enterprise Cloud Platforms.