Framework for Network Services Orchestration with a Cloud Native Approach

The management of modern telecommunications networks is complex, driving the adoption of Software Defined Networking (SDN). A typical SDN approach may rely on a monolithic architecture, with very strict resource requirements and a centralized database, that could lead to scalability issues in large-scale environments. This thesis faces these challenges by designing, implementing, and validating a new framework for network services orchestration based on cloud-native principles. The proposed architecture uses Git as a single source of truth and leverages GitOps workflow principles to translate high-level network service definitions into low-level device configurations, which are then applied to the targets using the gNMI protocol. Implementation required the development of a completely new module and the contribution to an open-source project, Kubenet SDCIO, that is in charge of the configurations provisioning to the network devices. The framework functionality was validated both on virtual and physical network devices. The successful deployment and positive performance results confirm the scalability of the proposed solution, also for large-scale networks. This work concludes that the adoption of cloud-native principles provides a robust and scalable solution for modern network automation.