Efficient restoration capacity design in MPLS networks

Abstract

In order to support stringent service level agreements, MPLS network design should consider both service and restoration capacity. There exist a range of restoration schemes in MPLS networks, such as routing protocol converging restoration and path-based LSP restoration. In path-based LSP restoration, each LSP is associated with two paths: a primary path and a restoration path, where restoration paths do not consume bandwidth until their primary paths fail. Thus, the restoration paths can share restoration capacity on common links as long as their primary paths do not fail at the same time. To achieve the maximum sharing of the restoration capacity, an efficient path selection algorithm is very crucial. We propose an efficient algorithm for selecting LSP restoration paths in MPLS networks. Our simulation on an intercity backbone network shows that the proposed algorithm achieves the best sharing compared with other two well-used algorithms. Specifically, our algorithm requires 10%-25% less restoration overbuild, which is the ratio of the total capacity for restoration to the total capacity for normal service only.