Packet routing has always been an issue that affects network performance. The traditional protocol is known to work with local information and thus not able to produce a global optimal decision. In the software-defined network (SDN), its centralized controller obtains unified access to the entire network topology information and has the ability to partially solve traditional network packet forwarding problems. SDN controller uses the Dijkstra algorithm to find the shortest path calculated from the source node to the destination node. However, constraints such as the need to bypass the node which has a high rate of failure exist to prevent the Dijkstra algorithm from meeting this demand. In practice, we need not only consider the shortest path but also consider the second short path, the third short path, and so on. K-shortest paths algorithm discovers a set of paths ordered in the most optimal, optimal, and suboptimal, has a very wide application is integrated into SDN controller to handle routing functionality. In this study, instead of only the number of hops, bandwidth and delay are adhered to within the k-shortest path to select the final route. Under this circumstance, different topologies are examined in response to network bandwidth and packet delay. The experiment shows that tree topology is best suited for improving bandwidth while simple topology for reducing delay.