Generic Intent-Driven Networking Paradigm with Different Levels of Policy Abstraction

Xianglin Liu, Tong Li, Chungang Yang, Ying Ouyang, Zhu Han, Mohsen Guizani

Abstract: Intent-driven networking automatically refines high-level intents into network configuration policies, thereby simplifying network management complexity, for a large-scale network in particular. The intent refinement technique facilitates the full-lifecycle refinement from declarative intents to executable policies, and finally to device-centric configurations. Currently, policies at different abstraction levels during the intent refinement process lack a generic framework. Therefore, in this article, we propose a novel network paradigm to present the hierarchical framework of policies and the full-lifecycle of intent. We survey existing policy languages and classify them based on different abstraction levels. In addition, we distinguish the intent refinement techniques according to template-based, model-based, and search-based design approaches, and then analyze refinement between different levels of policies. Finally, we design an intent refinement system to validate the feasibility and effectiveness of our proposed framework.