Design And Implementation Of Network Fault Diagnose System Based On Segment Routing

Network fault management is of great significance for network running smoothly,a good network fault management method can effectively reduce the network troubleshooting time and recovery time and increasing the working efficiency of the operation and maintenance engineers.The current popular fault localization techniques include passive monitoring techniques and active monitoring techniques.However,the Equal-cost Multi-path(ECMP)links in the network brings uncertain to these traditional fault localization techniques based on IP layer in diagnosis process while the actual nodes or links the probing packets going through are not sure.To avoid this problem,the network monitoring method based on Segment Routing architecture can be introduced,which can get accurate probing path and simplify fault diagnose process even effectively reduce the quantity of probes required.This article will design and implement a network fault detection system based on Segment Routing,which mainly solves the problem of fault detection and location in the network.First of all,it analyzes the functional requirements of the system,including functions of probe deployment and path construction,fault monitoring,fault location and probe dispatching,and determines the system development goals.After that we conducted a research on the existing algorithms in solving probe deployment and path building.In order to cover the network and reduce monitoring redundancy as much as possible,an improved algorithm SCLMon(SCMon with minimum monitoring cycle length)combined with centrality probe deployment strategy is proposed.First,the quantities of path covered links and adjacent uncovered links of target node are introduced as evaluating indicators in optimal monitoring path selection.In addition,the betweenness and the degree are utilized to assess the centrality of the nodes and the node with maximum centrality is picked for probe deployment.The experimental results show that SCLMon algorithm with centrality probe deployment strategy can effectively reduce monitoring redundant in different network topologies or at different maximum number of segments.Besides that,this paper proposes a multi-probe deployment and probing path generation algorithm based on single-probe deployment algorithm and tested it with real network topologies.The results show that the proposed algorithm can further reduce the monitoring redundancy compared to single-probe deployment algorithm.Based on the above improved algorithm we designed and implemented a network fault detection system,including the probe deployment and monitoring path generating module,fault monitoring module,fault localization module and probe dispatch module.The probe deployment and path building module is implemented based on the proposed SCLMon algorithm.The goal of module is to obtain the probe deployment and the monitoring path building plan.The fault monitoring module is used to detect anomaly in the network.The fault localization module is triggered by monitoring module and can localize fault by sending new probes to the network.The probe dispatch module is built for external probes to communicate with the system.Finally,we tested the system in a virtual environment.The results show that the system can get a reasonable deployment plan and monitoring paths in various scale topologies.It can also monitor network running status and trigger fault localization process when anomaly is found.In the case of simple fault,the system can detect and locate the fault effectively and accurately.