An efficient backup path selection algorithm in MPLS networks

Abstract

The rapid growth of real-time and multimedia traffic over IP networks makes not only QoS guarantees but also network survivability more critical. This paper proposes an efficient algorithm which supports end-to-end path-based connection restoration in MPLS networks. We review previous related work. This includes SPR (Shortest Path Restoration), PIR (Partial Information Restoration) and CIR (Complete Information Restoration). The objective of backup path selection algorithms is to minimize the total network bandwidth consumed due to backup paths. Backup path bandwidth usage can be reduced by sharing backup paths among disjoint service paths. In CIR, since a path selection algorithm uses per-LSP information, backup path sharing can be optimized. However, the large amount of information, which each node advertises and maintains, makes it impractical. In the case of PIR, some sharing of backup paths is possible while using the aggregated service bandwidth and backup bandwidth used on each link. We think it is reasonable to increase backup path sharing using aggregated information as with PIR. In this paper we propose an efficient backup path selection algorithm to outperform PIR while using aggregated information. Simulation results show that our algorithm uses less total backup bandwidth compared to PIR.