Efficient failure recovery techniques for segment-routed networks

In communication networks, protecting against node and link failures is an important requirement. In data networks such as IP/MPLS networks robust protection techniques have evolved, while also depending on the underlying transport networks to provide reliability. In networks such as data center networks, it is not possible to rely on the underlying layer for handling of failures. It is necessary to implement protection mechanisms with fast recovery times. Fast reroute is one such mechanism where, upon a failure, repair is initiated at the point of local repair (PLR). However, this is not efficient in terms of the operational cost of the repair. With segment-routed networks, an opportunity arises to initiate the repair at the segment endpoints and not the PLR. In this paper, we present a segment-level recovery framework, where protection is applied for each segment of an end-to-end path rather than individual links or nodes. The proposed technique is more efficient than two other existing techniques that use segment routing for recovery, namely Topology Independent Loop Free Alternate (TI-LFA) and Topology Independent Multi-Failure Alternate (TI-MFA). Simulation-based results show that the proposed scheme provides efficient failure recovery in terms of flow drop rates, ability to recover from multiple failures and resources required for the recovery. Based on these results we find that the drop rate of flows due to failure can be up to 62% lower and the increase in operational path length can be up to 35% lower resulting in more robust protection and significantly better resource utilization.

Automated summary

In this paper, a segment-level recovery framework is proposed for protection of each segment of an end-to-end path, which proves to be more efficient than other existing techniques using segment routing for recovery. Simulation results demonstrate that the proposed scheme offers efficient failure recovery in terms of flow drop rates, ability to recover from multiple failures, and required resources for recovery, resulting in lower flow drop rates and path length increase, leading to robust protection and better resource utilization.