Abstract: Virtualization technology can greatly improve the efficiency of the networks by allowing the virtual optical networks to share the resources of the physical networks. However, it will face some challenges, such as finding the efficient strategies for virtual optical nodes mapping, virtual links mapping and spectrum assignment. To tackle these challenges, a constrained optimization model with the number of frequency slots and the maximum index of used frequency slots minimized was established to determine the optimal virtual node mapping, virtual link mapping and routing schemes. For the sake of solving the model efficiently, tailor-made crossover, mutation and local search operators were designed. Based on these, an efficient genetic algorithm was proposed. To test the model and the proposed algorithm, the simulation experiments were conducted on two widely used networks, and the experimental results showed the effectiveness of the proposed model and algorithm.