Generic Controller Adaptive Load Balancing (GCALB) for SDN Networks

Abstract

Fault tolerance is an important aspect of network resilience. Fault-tolerance mechanisms are required to ensure high availability and high reliability in different environments. The beginning of software-defined networking (SDN) has both presented new challenges and opened a new era to develop new strategies, standards, and architectures to support fault tolerance. In this paper, a study of fault tolerance is performed for two architectures: (1) a single master with multiple slave controllers and (2) multiple slave controllers. The proposed model is called a Generic Controller Adaptive Load Balancing (GCALB) model for SDNs. GCALB adapts the load among slave controllers based on a GCALB algorithm. Mininet simulation tool is utilized for the experimentation phase. Controllers are implemented using floodlights. Experiment results were conducted using GCALB when master controller is taking the responsibility of distributing switches among four and five slave controllers as a case study. Throughput and response time metrics are used to measure performance. GCALB is compared with two reference algorithms: (1) HyperFlow (Kreutz et al., 2012), and (2) Enhanced Controller Fault Tolerant (ECFT) (Aly and Al-anazi, 2018). Results are promising as the performance of GCALB increased by 15% and 12% when compared to HyperFlow and by 13% and 10% when compared to ECFT in terms of throughput and response time.