Member： Thomas KRENC

Towards Next-Generation Route Analysis Platforms

Abstract

Internet routing data is highly complex and requires specialized knowledge and significant time to process and analyze. This work presents a novel and efficient method for network operators to troubleshoot their networks without directly interacting with routing data.

Introduction

The Internet is a complex and dynamic system where a wide range of things can go wrong — from configuration errors to policy conflicts and routing attacks. Network operators rely on timely insights into routing from complex routing data to troubleshoot issues, detect anomalies, and validate policy changes. However, existing route analysis platforms, such as BGP.tools and RIPEstat, lack the flexibility to accommodate the diverse and evolving set of questions operators need to ask about network state.

The Next-Generation Route Analysis Platform is a framework that enables network operators to query specific networks using natural language, without requiring prior expertise in complex routing data. It includes an LLM component that interprets human-formulated questions to extract their context. Based on this context, the system initiates a targeted data transformation process to generate the information needed to answer the query in a scalable and efficient manner.