MULTI-OBJECTIVE RESOURCE SCHEDULING FOR PERFORMANCE AND CARBON EFFICIENCY IN KUBERNETES

Abstract

Kubernetes has become the leading container orchestration platform for cloud-native enterprise applications because it enables scalable resource management, automated workload scheduling, and high application availability. However, conventional Kubernetes schedulers primarily optimize computational performance and resource utilization without explicitly considering carbon emissions and energy efficiency. As enterprise cloud infrastructures continue to expand, balancing application performance with environmental sustainability has become an important operational challenge. This paper proposes a multi-objective resource scheduling framework for Kubernetes by integrating machine learning, predictive analytics, cloud-native monitoring, carbon-aware computing, and intelligent resource orchestration. The proposed framework continuously evaluates workload characteristics, cluster utilization, application priorities, renewable energy availability, carbon intensity, and Quality of Service (QoS) requirements to generate optimal scheduling decisions. Experimental analysis demonstrates significant improvements in workload performance, infrastructure utilization, energy efficiency, carbon emission reduction, and sustainable cloud operations compared with conventional Kubernetes scheduling approaches. Keywords: Kubernetes, Multi-Objective Scheduling, Carbon Efficiency, Machine Learning, Predictive Analytics, Cloud-Native Computing, Green Computing, Resource Optimization, Sustainable Cloud, Container Orchestration.