RSAFE: A Robust Software-based Fault-Tolerant Scheme for Large-scale Ethernet Networks

Most conventional software-based approaches for fault tolerant Ethernet (FTE) adopt a heartbeat mechanism for fault detection. However, due to the tradeoff between the number of nodes and the fault detection time, these conventional approaches do not provide scalability for large-scale networks. To solve this scalability issue, we previously proposed the SAFE scheme, which utilizes multiple subnets architecture. In this paper, we present an upgraded version of the SAFE scheme, the rapid SAFE (RSAFE) scheme, which adopts novel algorithms to significantly decrease the fault detection time while maintaining the advantages of the SAFE scheme. Simulation results using OMNeT++ show that the RSAFE scheme achieves excellent fault detection performance compared to the SAFE scheme. Furthermore, new analytical derivations and discussions of the key advantages of both the SAFE and RSAFE schemes are presented based on the multiple subnets architecture.