A Study on Dynamic Survivable Routing with Availability Constraint for GMPLS-Based Recovery

This paper introduces a new dynamic availability-aware survivable routing scheme under the framework of generalized multi-protocol label switching (GMPLS)-based recovery, which aims to achieve the best generality for the network operation in meeting the end-to-end (E2E) availability requirement of each connection. The paper first defines the partial restorability, justifies the feasibility of equipping a connection with partial restorability under the GMPLS control plane, and highlights the approach of evaluating the E2E availability for a pair of working and partially restorative SRG (shared risk group)-disjoint shared backup label switched paths (LSPs). A compact matrix expression is developed for modeling the minimum spare capacity along each link and the cost function for solving both of the paths. Based on the developed cost function, a novel integer linear program (ILP) is formulated to dynamically determine the working and backup LSPs of the corresponding connection request with the least amount of total capacity while the E2E availability requirement of the connection is met. We demonstrate that the model is general to the traffic uniformity, connection indivisibility, and working bandwidth restorability compared with the previous studies. Simulation is conducted to verify the proposed ILP model by making a comparison with a number of legacy schemes that achieve 100% restorability in facing a specific number of simultaneous failures, such as shared path protection (SPP), 1+1 protection, and dual-failure protection. In the case study, we have seen merits in the proposed algorithm by significantly reducing the required redundancy in the effort of achieving the given availability constraint for each connection request.