Software-defined networking (SDN) is a network technology that explicitly demarcates network control tasks from forwarding tasks. The infrastructure and the control plane are separated. It centrally specifies the control logic and manages the whole network. The network is controlled by a software program called a controller. A single controller is not sufficient to deal with the problem of scalability and dependability. To solve these problems, more number of controllers are required. Placing various controllers is called controller placement problem (CPP) in SDN. The problem is to select the count and location of controllers to be placed in the network. Control latency is an important metric in SDN as all functions are carried out through frequent message exchanges between the switches and controllers. Latency comprises mainly the processing, queuing, propagation and transmission latency. In a wide area network, propagation latency is an important metric which includes the average and worst case node to controller propagation latency. Besides, inter controller latency plays a significant role to maintain consistency in large networks. The sum total of switch-to-controller and inter-controller delay is known as the global latency. The objective of this paper is to improve SDN performance by lowering the network's overall latency in an optimal manner. A comparison of optimal with the random approach is shown in the results. Finding controller number while minimizing latency is another objective of this paper.