LLM-Based AI Agent for VNF Deployment in OpenStack Environment
Sukhyun Nam, Nguyen Van Tu, James Won‐Ki Hong · 2025
This paper presents a novel approach to automating the deployment of Virtual Network Functions (VNFs) in an OpenStack environment using Large Language Models (LLMs). Building on the concept of Intent-Driven Networking (IDN), which allows network administrators to manage complex networks via natural language commands, we explore the feasibility of using LLMs to automate VNF deployment tasks. A dataset of Method of Procedure (MOP) documents was created and utilized to prompt LLMs to generate Python code for deploying and configuring VNFs. Our LLM-based AI agent framework tests the generated code within an OpenStack environment, comparing the performance of various LLMs. Our findings highlight both the potential and current challenges of using LLMs in network automation, suggesting pathways for future research, including advanced prompt engineering and real-time error correction.