In an extensive quickly growing DWDM transport network the frequency of in-service upgrades or other required maintenance activity may be high. We are therefore interested in defining, quantifying and managing the theoretical risk that a "maintenance outage" may have for the rest of a mesh-restorable network. The point of view is of putting one span in maintenance and then asking, if a failure occurs to another span while in the maintenance state, what is the risk of incomplete restoration? In ring networks, the corresponding exposure to failure is contained within the specific ring involved. But the price for this containment is the 100% redundancy of rings and that the risk is of 100% restorability loss on other spans when in a maintenance state. We find that mesh networks exhibit a more distributed risk field, but much lower risk magnitudes. In over 2500 trials with a mesh equivalent of roll-to-protection as the maintenance model no other span sustained over 70% theoretical loss of restorability and 99.6% of the time the theoretical risk to restorability was < 50%. This gives an important operational advantage in support of priority service paths. We give a procedure for calculating the 'theoretical risk field' of a given maintenance action in a mesh-restorable network. We then study the effects of various mesh-network design differences on the extent and magnitude of the theoretical risk field. We produce results for a worst-case ("equivalent to failure") model for maintenance and the mesh equivalent of "roll to protection" in rings. The basic methods can be adapted for other maintenance models or models of the restoration process or adapted to handle multiple maintenance scenarios. The work is aimed at understanding and managing the effects of maintenance in a mesh-restorable network and ultimately providing new operational and design capabilities for future mesh-based network planning and operations systems.
展开▼
