Efficient Bandwidth Allocation for Backup Paths

In MPLS networks, failure resilience of a Label Switched Path (LSP) is essential in guaranteeing the QoS of the path. This can be achieved by allocating backup LSPs through which packets can be routed in case of primary LSP failiures. bandwidth efficiency and expeditious path restoration are two desirable features for the design of backup LSPs. This paper tackles the bandwidth efficiency with a scalable and robust algorithm for backup path bandwidth sharing based on statistical aggregation. By analyzing the relationship among the amount of allocated backup bandwidth, the number of backup paths, and path failure probability, given an upper bound of fatal failure probability, we find the amount of backup path bandwidth can be approximated by the product of the average primary path bandwidth and a quadratic function of the number of primary paths and the average primary path failure probability. Simulation results show that the fatal failure probability is less than 0.5% as the number of primary paths spans from 1 to 110, the primary path failure probability is uniformly distributed between 0.001 and 0.2, and the primary path bandwidth is uniformly distributed. Compared to the generic approach of allocating a backup path with the same bandwidth as the primary path, this algorithm is 50% to 80% more effcient. Compared to the baseline case where no backup bandwidth is allocated, the capacity loss for maintaining restorable primary paths is only 20%. Keywords—Backup Path Recovery, MPLS