Dynamic Resource Allocation for Cloud-Native Network Functions Based on Energy Efficiency

Telecom network services are transforming to cloud-native network functions with greater flexibility and faster scaling capabilities. However, this enhanced flexibility brings new challenges for resource allocation and energy efficiency, especially in large-scale CNF deployments. We proposed a dynamic resource allocation algorithm based on Kubernetes pod autoscaling to optimize CNF utilization while maximizing energy efficiency. In addition, CNFs are scaled efficiently by adopting approaches integrating energy and performance monitoring, event-driven scaling, and resource optimization. According to the pilot site results, the dynamic resource allocation approach enhances the CNF energy efficiency ratio and increases the power consumption of worker nodes. The CNF energy efficiency ratio increased from an initial $83.9 \%$ to $90 \%$ with the dynamic resource allocation approach.