Numerous Internet of Things (IoT) devices are being connected to the networks to offer services. To cope with a large diversity and number of IoT services, operators must meet those needs with a more flexible and efficient network architecture. Network slicing in 5G promises a feasible solution for this issue with network virtualization and programmability enabled by NFV (Network Functions Virtualization). In this research, we use virtualized IoT platforms as the Virtual Network Functions (VNFs) and customize network slices enabled by NFV with different QoS to support various kinds of IoT services for their best performance. We construct three different slicing systems including: 1) a single slice system, 2) a multiple customized slices system and 3) a single but scalable network slice system to support IoT services. Our objective is to compare and evaluate these three systems in terms of their throughput, average response time and CPU utilization in order to identify the best system design. Validated with our experiments, the performance of the multiple slicing system is better than those of the single slice systems whether it is equipped with scalability or not.