Recently, many researchers have tried to answer a question regarding smart cities: how to connect a wide variety of sensors and aggregate the data collected by different technologies related to the Internet of Things (IoT)? Those data are usually gathered and transmitted through WAN (Wide Area Network) or private MPLS (Multi-Protocol Label Switching) connections. However, since WAN based solutions present different resources in terms of data, size, coverage area, latency and capacity requirements, they become inefficient or even prohibitive regarding operating costs on smart city applications. An alternative is the use of SD-WAN (Software-Defined Wide Area Network), which combines hardware and software appliances or is software based only, consisting of the virtualization of WAN connections. The main characteristics of SD-WAN are: the ability to do dynamic path selection, facilitating data flow and increasing system resilience; the support to multiple connection types (ADSL, VDSL, FTTH or 3G/4G), enlarging the coverage area when compared with traditional WAN; the employment of a simple interface (easy to configure and manage); the capital and operation expenditure reduction; the increase of service agility and flexibility; the implementation of a centralized control and monitoring with lower costs. In this sense, this paper proposes an SD-WAN embedded white-label solution, of low cost and low energy consumption for commercial and academic use employing Ubuntu Linux, Floodlight SDN Controller, Open vSwitch and Raspberry Pi 3b. Then, the Quality of Service (QoS) parameters obtained by the proposed SD-WAN solution are measured and evaluated, through network emulation tests performed with the aid of Distributed Internet Traffic Generator (D-ITG), with the purpose of verifying the applicability in smart cities.