Hybrid SDN-NFV orchestration for scalable network function deployment in cloud-native data centers

Abstract

Hybrid cloud-native data centers supporting 5G, IoT, and latency-sensitive edge applications increasingly rely on software-defined networking (SDN), network function virtualization (NFV), and containerized cloud-native network functions (CNFs), yet existing orchestration mechanisms remain fragmented across disparate control stacks. This study proposes, implements, and evaluates a hybrid SDN-NFV orchestration framework that unifies ETSI NFV-MANO lifecycle management for VNFs/PNFs, Kubernetes-based orchestration for CNFs, and SDN control of data center networking under a single intent-driven service graph. Using a multi-tenant testbed with mixed VM-based VNFs, CNFs, and PNFs, we assess performance across three orchestration modes: baseline loosely coupled control, NFV-centric orchestration, and the proposed hybrid approach. Statistical analysis of 360 instantiation experiments and extensive load testing reveal that the hybrid framework reduces service instantiation time by more than 50% compared with baseline configurations, achieves significantly higher CPU/memory utilization and network function density without SLA degradation, and maintains superior latency and packet-loss compliance under heavy load. Moreover, the hybrid design demonstrates improved control-plane scalability, with lower SDN controller CPU consumption, higher sustainable instantiation rates, and faster failure recovery as tenant counts increase. These results highlight the operational value of integrating VNFs, CNFs, and SDN connectivity into a unified orchestration layer, offering a scalable architectural model for cloud-native data centers transitioning toward fully software-defined 5G and edge services. The study concludes that hybrid SDN-NFV orchestration can address key limitations of existing siloed orchestration systems and provides actionable guidance for operators and architects seeking to modernize network function deployment in heterogeneous, multi-tenant environments.