Local Fast Reroute With Fine-Grained Monitoring and Priority-Aware Congestion Control

Abstract

Communication links or nodes failure is common in Software-Defined Networking (SDN), which affect on-going communication and induces performance bottleneck. In SDN, protection-based recovery mechanisms pre-installs a backup route to offer local failure recovery. In those mechanisms, rule compression improves memory consumption. However, rule compression reduces network visibility and impedes fine-grained monitoring. When a topology has edge-disjoint routes, protection-based recovery uses OpenFlow Fast Failover Group (FFG); otherwise, it uses MPLS-crankback (CB) to reroute the affected traffic. But CB strategy suffers from continuous backtracking and induces extra latency overhead. Crankback with Controller (CBC) strategy can terminate such backtracking. CBC also suffers from the controller to switch communication overhead and impact the overall recovery time. Moreover, congestion in a network induces packet loss and impedes reliable communication. Priority flows can experience longer routes because of load distribution to alleviate congestion.