A system and method for implementing a shared protection scheme under a scenario of multiple failures in an optical network. In one embodiment of the invention, a working path between an ingress node and an egress node is computed responsive to a connection request received in the network. One or more backup paths are computed between the ingress and egress nodes, which are activatable upon a failure condition associated with the working path or the backup paths. The backup paths may be based on link and/or node disjointedness, as well as resource-based cost constraints.