Architecting Autonomous Network Management and Control via Intent-Driven Decoupled Network

Abstract

The current coupled network management architecture and control protocols restrict the network flexibility and scalability. In order to achieve autonomous network management and control, it is necessary to decouple network functions across various layers, including application, control, data, and management layers. Although there exists emerging decouple concepts like Intent-driven Network, Software-defined Network, Kubernetes, and Network Function Virtualization, there is a lack of a generic network management architecture to achieve a full-lifecycle autonomous network management. To fill the gap, we design a generic decoupled network management architecture and interface that achieves higher flexibility and scalability by exploring and exploiting potentials of decoupled network management components. Given the decoupled network architecture, we present an intent-driven autonomous network management and control scheme, known as SAI, considering the network state, action, and potential intent. Finally, we build the intent-driven data network management prototype to prove the concept and evaluate the performance of the presented SAI management and control scheme.