How Latency Meets its Match with Network Slicing

For many service providers, latency is a killer. Imagine trying to use Artificial/Virtual Reality (A/VR) and having your experience interrupted by jittery images with uneven sound synchronization. That wouldn’t exactly be “reality” would it? Or maybe you’re online gaming and you have a target in your sights, but before you can get a shot off bang, you’re dead from an unseen enemy with a little less lag than you. That latency isn’t going to gain too many fans of your service.

Latency is highly dependant on distance, about 5 microseconds per kilometer the traffic travels. From a service provider perspective, the packet core is very centralized in most networks today. The inbound and outbound traffic must travel all the way to where the packet core resides until it can be broken down into a service. That’s why it’s important to push the user plane deeper into the network and as close as possible to the subscriber while keeping the control plane centralized.

The application must be capable of reaching the end-user quickly enough to prevent the degradation of the experience. Network slicing and segment routing provide intelligent routing and traffic differentiation required to efficiently support this distributed architecture.

What is Network Slicing?

Network slicing is a flexible, scalable architecture that allows the multiplexing of virtualized, independent networks on the same physical infrastructure, taking advantage of concepts such as Software Defined Networking (SDN) and Network Function Virtualization (NFV). It enables the management of multiple logical networks as virtually independent business operations on a common physical infrastructure.

It offers the ability to partition mobile networks into a set of virtual resources, and each “slice” can then be allocated for different purposes. For example, it enables service providers to route latency-sensitive traffic such as A/VR and gaming into slices, while routing less latency-sensitive traffic such as video streaming differently so those services are not affected. It is a key concept in 5G and a way to utilize the network in a more intelligent and cost-effective way than ever before.

Some other examples of network slicing utilization might include allocating a slice to a Mobile Virtual Network Operator (MNVO), an Internet of Things (IoT) domain, an enterprise customer, or another set of services such as mobility as a service. A network slice also extends the Access Point Name (APN) concept used in today’s mobile networks.

From the transport network perspective, we envision network slicing starting from the cell-site router, as close as we can get to the radio. In this way, we ensure the Service Level Agreement (SLA) runs end to end from the cell site to the packet core living in the data center. The transport slice through segment routing forms the basis of our comprehensive xHaul portfolio.

How Does Network Slicing Differ from Segment Routing?

Network slicing and segment routing are two separate functions that work together to improve the end-user experience. Our engineers developed segment routing in 2013 as a flexible and scalable method of performing source routing, where the source selects a path and encodes it in the packet header as an ordered, prioritized list of segments. Segments are an identifier for any type of instruction. Each segment is identified by the Segment Identifier (SID) consisting of a label, unsigned 32-bit integer.

Segment routing is gaining popularity as a means of simplifying Multi-Protocol Label Switching (MPLS) networks. We see segment routing changing the way MPLS networks function and facilitating the adoption of SDN. Because segment routing directs traffic on a stateless, flexibly defined path, it has the benefit of being programmed by an SDN controller or locally by the head-end source-based routing.

This solution actively monitors latency, and if there is any change due to higher congestion, for example, the traffic is automatically rerouted without the need for human oversight to manually configure the change. Since it is fully stateless it is higly scalable and we can have many SLAs across secure networks. And we can do that across multi access types with wired, wireless, and business services all converged on the network. Used with network slicing, segment routing offers the ability to logically instantiate a private network service within the larger service provider network. It helps enable service providers to deliver secure, high-speed, low-latency content to their customers. With segment routing, you can assure the best management of traffic flows, enabling any-to-any connectivity with multiservice support, smooth integration with the data center slicing, and all that in a scalable and simpler way.

So What’s in it for Me?

Because network slicing is the instantiation of a logical network or function, it can be managed separately from the rest of the logical and physical network. While you can instantiate a network slice only within the packet core, or the Radio Access Network (RAN), the larger goal is to provide a logical end-to-end, cloud-to-client slice.

The physical equivalent of this is an operator who builds a private network designed for a specific offering or customer. Naturally, doing this in a logical rather than physical manner is more agile, flexible, scalable, and less expensive.

In the examples cited earlier, there are obviously several market subsets for service providers that are more likely to pay a premium for latency-free services, such as gamers and A/VR users. Therefore, network slicing can clearly be monetized by service providers.

A lot of what we see in the market is limited to a small number of slices and it’s suboptimal from a utilization perspective of the investment. For example, some of our competitors offer to slice, but it’s considered “hard” slicing where you might have a very limited number of slices in the network, and different applications get put into those limited slices. It’s also very static.

You want to get the most utilization of your network, but those capabilities are too limited to accomplish that. We’ve made our network slicing both profitable and efficient, and because it’s virtualized it can fit very nicely on our Telco Cloud infrastructure.

Learn More

5G is changing the way we build and utilize networks, so for service providers, this is both a challenging and exciting time. Explore all of our mass-scale network transformation solutions including network slicing, segment routing, xHaul, and our Telco Cloud Infrastructure, and see which solutions are right for your operations.