5G is being envisioned as a network that can be manipulated on the fly to accommodate a number of different scenarios, or use cases. But to make that vision a reality, network operators will have to be able to orchestrate certain capabilities across different points of the network — a phenomenon that is called dynamic network slicing.

Network slicing capitalizes on the capabilities of software-defined networking (SDN), network functions virtualization (NFV), orchestration, and analytics.

In 5G mobile networks, network slicing allows operators to design, deploy, and customize different “slices” of the network, running on a common network infrastructure.

One reason vendors like Nokia are calling it “dynamic” network slicing is because operators will be able to designate these network slices quickly. In fact, one use case for dynamic network slicing that Nokia describes is in a stadium where network traffic escalates before the game starts but it’s primarily on the downlink because event-goers are downloading maps or getting information about the event that is happening in the stadium. Once the game or event begins, the traffic shifts to the uplink because people are uploading video and photos to social media. “We envision a scenario where an operator could tailor the network based upon the traffic pattern they see in any environment,” says Jason Elliott, 5G market development manager at Nokia.

While 4G networks today can segment parts of the network to provide quality of service (Quality-Of-Service (QOS)) for certain services and applications such as Voice over LTE (VoLTE), some argue that VoLTE is not an example of network slicing. According to Elliott, VoLTE is made possible through prioritizing traffic rather than slicing network assets. “Network slicing is not implemented and has not been used before,” Elliott says. “It’s definitely not used in networks today as it is being proposed for 5G.”

Likewise, Peter Linder, head of business management for cloud and IP at Ericsson, says that virtual private networking (VPN) isn’t a good example of network slicing either because while operators are orchestrating certain network functions, they have to do it in a rigid way – meaning that it has a lot of limitations.

However, Linder adds that network slicing is possible with 4G, as network operators virtualize their network. “The fundamental idea of network slicing is possible in 4G and will come as we virtualize the 4G network,” Linder says.

He adds that the main difference between network slicing over 4G versus over 5G is that with 5G network slicing will incorporate the radio. “In 4G we don’t distinguish between what the device is at the other end.” But with 5G, operators will need to consider the end device. For example, if it’s some sort or machine-to-machine (M2M) application where battery life of the device has to last to six months, an operator will need to optimize the network for a low power application that doesn’t drain the battery.

Ericsson is working with SK Telecom to bring network slicing to the 5G core. The companies announced a collaboration last year that would ensconce an element of SDN inside next-generation mobile networks.

Of course, for mobile operators to be able to slice the network assets dynamically, they will have to rely upon analytics. Elliott says that network analytics will be able to model traffic patterns for operators so they can predict patterns. “You have to understand network performance. Analytics is about understanding and predicting how this resource might be used,” he says.

However, Ericsson’s Linder warns that network slicing will not be applicable to every application. “Some things will just need a good pipe,” he says.