The invention discloses a
parallel routing optimization
algorithm under an elastic optical network, and the
algorithm comprises the steps: step 1, after a batch of services arrive at the network, using a GPU to solve a shortest path on an available hierarchical graph of each service as an alternative path; step 2, after calculating an alternative path, selecting a
route for the service by using aroute
selection algorithm; step 3, judging whether the services can be added into the network, if so, ending the
algorithm, and if not, performing the step 4; step 4, judging whether the services which are not added are blocked services or not, if yes, ending an algorithm, outputting a path selected for the services, if not, re-calculating a
route for the services which are not added, and returning to the step 1. According to the method, the powerful
parallel computing capability of the GPU is fully utilized, and the computing speed of the routing optimization algorithm is increased.