Network virtualization is becoming a promising way of removing the inherent ossification of the Internet, and has been steadily attracting more and more researchers’ attention during the decades. The major challenges in this field are improving the efficiency of virtual network embedding procedure and raising the rate of virtual network requests being successfully accepted by the substrate network. This paper introduces a new virtual network embedding algorithm based on node similarity, which means the similarity between the virtual nodes and the substrate nodes. For more details, by calculating the degree of nodes both in virtual network and substrate network, which is actually the number of links associated with them, the algorithm achieves better mapping results between virtual network and the substrate network on the topology aspect.