Autonomous network monitoring using LLMs and multi-agent systems

This study researches the combination of the concept of large language models (LLMs) and multi-agent systems implementation into autonomous network monitoring. Dedicated to real-time anomaly detection in network telemetry (NetFlow, SNMP, GNMI), the study analyzes the possibility of GPT-based agents in efficiently identifying and responding to a network problem. The study also explores some details of Lang Graph and Auto Gen in construction of multi-agent systems to triage and remediate the network events. A case study on automating Root Cause Analysis (RCA) in spine-leaf topologies illustrates the practical application of these technologies. The results indicate that the detection of anomaly is quite efficient in speed, accuracy, and scalability and serve as a rather efficient way of simplifying network operations. The paper highlights, as the recognition of the role of AI in NetOps increases, the potential of converting network management with enhancements to automation and decision-making processes that LLMs and multi-agent systems provide. The implications are huge in terms of the future of NetOps where network infrastructures can be smarter and self-healing.