Load Aware Controller Failover and Controller Restoration Mechanisms in SDN Control Plane Using Rest-API

Abstract

Single point of controller failure is a crucial matter of single controller SDN applications. In addition to the vulnerability of single point of failure, single controller applications face low-scalability and performance problems. Due to these issues, multiple SDN controller applications are propounded. Multiple SDN controller applications offer scalability, fault tolerance, and consistency. To enable fault tolerance feature, multi-SDN controller paradigm necessitates controller failover methods. Even though the aim is failure tolerance, the methods differ based on the considerations of failover. In this study, controller load is chosen as failover consideration. First, we offer a load monitoring application leveraging JSON objects provided by RESTful servers. The load monitoring application is used for a non-redundant controller failover process. In addition to load-aware failover mechanism, a controller restoration procedure is also proposed in case of a failed controller existence or a false failure detection. REST API is used for both failed and restored controller detection.