A Framework for Automatic Root Cause Analysis of Network Failure on MPLS Network

The topology and connectivity of Multiprotocol Label Switching Virtual Private Networks (MPLS VPNs) are highly complex, making it challenging to investigate and verify problems. Investigating and verifying issues within MPLS VPNs poses a significant challenge, especially concerning chain failures, where a single issue can trigger a cascade of failures across the network. Typically, the analysis process is carried out by experienced users who utilize event or alarm messages from network devices or the Network Management System (NMS). However, these messages often necessitate additional information obtained by executing operation and maintenance commands from network devices. This paper aims to develop an Automatic Root Cause Analysis system for MPLS VPNs use- cases, comprising different components: a Message Broker which supports real-time streaming and multiple sources of message types, a Log Template Generating service for creating generic log templates, an Event Identification Service for classifying types of event messages, a Nodes Chain Lookup Service for finding related affected network devices, a Node Test Service for executing commands to verify various network metrics, a Case-based Fault Identification Service serving as the database of expert knowledge or historical fault cases, a Notification Service for notifying through various modern channels, and a Dashboard providing information to network operators about the root cause of network failure and how to handle it. For evaluation, we utilized simulated generated Critical and Major event messages. This evaluation demonstrates the efficiency of the proposed system in terms of automation and speed.