The Internet of Things (IoT) refers to variety of smart devices such as smartphones, tablets, and sensors that can interact and exchange of data among devices through the Internet. The diversity of IoT devices and their services have posed a larger range requirements of availability, throughput, latency, and performance in heterogeneous connectivity environments. Meanwhile, the existing networks often struggle with such of limitations in complex control protocols and difficulty in internetworking with billions of smart devices with different requirements such as latency and bandwidth allocations. These obstacles become substantial barriers to deploy services, as well as isolate between multiple co-existing tenants on the same physical network, deploy simultaneous protocols in the network, be stable to maintain the bandwidth and latency according to predefined QoS demands. These obstacles have recently been facilitated by Software Defined Network (SDN) and Network Function Virtualization (NFV) technologies that enable the programming and monitoring in data plane. In this study, firstly, the authors investigate and propose a SDN/NFV based architecture for multi-tenant networks with plenty of network slices working in a shared physical infrastructure. Secondly, P4 and ONOS Controller are used to implement a deep programming in BMv2 devices to efficiently maintain the network motoring in order to guarantee the E2E latency of communicating channels. Finally, the VXLAN technologies are exploited to for network slicing with different purposes and applications, and Inband Network Telemetry (INT) is used to monitor network latency.