In the near future, the services provided by 5G will benefit many industries, with 5G IoT connecting billions of devices including sensors, cameras, wearables, vehicles, and even robots. 5G's ultra-reliable low-latency communication will unlock transformative updates in critical areas such as infrastructure or manufacturing, from the healthcare industry to the Internet of Vehicles (IoV), and from smart homes to smart cities. Can 5G really deliver these capabilities? Yes, but it requires a specific analysis based on the characteristics of each application, as different applications have their own requirements. Video transmission may require enormous downlink bandwidth but does not necessarily need low latency or high reliability. 

The IoV may require low latency and ultra-high reliability but does not need much bandwidth. Therefore, even if 5G technology cannot simultaneously achieve large bandwidth, low latency, and high reliability, it can still support each targeted application. To this end, 5G defines network slicing technology. What is network slicing? Network slicing divides a physical network into multiple virtual end-to-end networks that all share the same underlying resources but are enhanced for different applications. 

While completely physically isolated computing and network resources are feasible, they are not cost-effective, so network slicing uses logical isolation. In the past, logical isolation merely meant separating different user traffic. For example, a VLAN (Virtual Local Area Network) is a virtual instance of a LAN that ensures flooding only occurs within the VLAN in the case of Ethernet frame flooding. That's all. There is no resource isolation, so if one VLAN causes switch congestion, other VLANs passing through the same switch will also encounter this issue. Therefore, isolation is actually a best-effort isolation, although service providers have never admitted this to customers. Compared to VLANs, network slices are truly independent network resources. By using Network Function Virtualization (NFV), each slice has its own forwarding, control, and management entities (sharing slice orchestration and classifiers).

 Similarly, through the use of NFV, network slicing can achieve truly independent network fault management and performance segmentation, where the behavior of one slice has no impact on the performance of other slices. Although a physical network element failure may affect multiple slices, in extreme cases, a specific slice may require dedicated physical hardware (i.e., true physical isolation), but such situations are extremely rare.