HORIZONTAL NETWORK SLICING
HORIZONTAL NETWORK SLICING
Horizontal Slicing promises to create the foundation for the evolving long tail of slice-as-a-service business models. Here we look at how horizontal slicing differs from vertical and how this approach applies to the Application-Aware network concept.
Vertical Network Slicing focuses on serving specific Industry verticals or markets en masse. The first and most obvious example of this came when MNOs allowed virtual operators to use and sell their capacity to different customers. In the evolved Network Slicing paradigm, vertical slices will be many and driven by industry specific demand and requirements. But before going into the weeds on vertical slicing, we’ll take a look Network Slicing in all all of its flavours to get a handle the different elements at play.
At a basic level, network slicing can be understood along three axes: The first relates to the commercial purpose of a given slice as to the ‘Vertical’ industries or ‘Horizontal’ user experiences that they are designed to serve; the second, ‘Static’ or ‘Dynamic’ describes the manner in which they serve their function. This simply relates to whether or not the slices are pre-deployed, or created and delivered in a dynamic, real-time context based on human or machine interaction; the third becomes very technical very quickly but describes the specific locations within the network where the slices may be deployed that include the Core Network (CN), the Radio Access Network (RAN) or at the Resource Level.
While all of these flavors of slicing are available today, their happy coexistence within the context of a true End-to-End Vertical & Horizontal Dynamically Sliced Network (or an E2EVHDSN; What a mouthful!), is the stuff of 5G. But it’s important to note, and as we will illustrate in the coming section on standardization, that the pathway to that certainty is paved in 4G. “Network Slicing does work in 4G by the way!” So said Intel VP & GM for the 5G Infrastructure Division Caroline Chan of the Network Platform Group at Intel at the recent 5G North America conference in Austin. This is an important point, not merely because it points to the right technology path to 5G (as confirmed by 3GPP), but indeed, and as evidenced by the Cloudstreet/Telia use case, it provides the opportunity for new, 5G-type revenues today. Thus we look to 4G/5G network slicing PoCs as the currency of this evolution, and a clear and present pathway to revenue and ROI. To continue unpacking the concept, we will in the next section provide a basic description of the difference between vertical and horizontal slicing and what happens when the latter is deployed as a dynamic, real-time automated system.
Vertical network slicing differs from horizontal in that it seeks to serve vertical (generally industry) segments, while horizontal serves individual users or machines. The first examples of network slicing were vertical, and designed to serve specific industries with a set of specific requirements from the network. We can imagine these, similarly to any collective of Network Slices, as multiple networks operating upon a single network but they aren’t necessarily logical or virtual. In this case we are distinguishing not just horizontal and vertical network slicing, but indeed physical and virtual. It is perhaps no surprise that this approach became the first example of the technology, and again no surprise, it happened on the wireline side of things. Today, this framework is already well in play and there are quite literally too many examples to name. When it comes to intelligently distributing – and indeed differentiating – bandwidth and processing power between distributed network clients and the cloud, these operative concepts are indeed well-established. In the graphic below we can see the interplay between a simple vertically sliced network and the manner in which horizontal slicing, both static and dynamic provides the “brains” within the model.
Here we see a number of use cases within multiple categories (“Life-Critical”, “Mission-Critical”, “Media & Entertainment” Etc.) all of them being delivered appropriate combinations of static or dynamic slices based on capacity or delay sensitivity or a combination of both. At the bottom of the pyramid, which also generally represents the scale of the market, we see a set of miscellaneous differentiated service categories including an ad-sponsored, “Zero Rating” which here we have as the only use case being delivered “Best Effort” mobile connectivity. Today, nearly every large technology company maintains one or more private networks that deliver SLA-assured throughout and a certain, pre-defined measure of capacity and ultra-low latency from a commercial ISP. One need only think of today’s Content Delivery Networks like Akamai, and their provisioning technologies like FastTCP to help their business customers sell video subscription services to end users to see the point. Slicing in this context is already very much core of today’s cloud Internet DNA.
THE USER-DEFINED NETWORK: NETWORK SLICING IN PRACTICE
Network Slicing promises to radically transform the mobile, digital landscape but while the technology discussion is in full swing, what about the users who will experience it? And how about bringing those experiences to market and across the bridge to 5G?
On the mobile side however, there are many fewer examples of network slicing, whether vertical or horizontal, but all of them rely to one extent or another on the idea of parsing network capacity. In the mobile context, the vertical form executes itself by either segregating groups of SIM-enabled devices or by reserving radio network bands of a given leased frequency with the intention of re-selling them to different customer segments. This manner of slicing is the basis of the telecom business model that allows Mobile Virtual Network Operators (MVNOs) to share resources with traditional operators and serve specific vertical customer segments.
Nonetheless, and because big or small radio antennae deployments do nothing to guarantee reliability, the mobile bottleneck remains and user experience suffers. Though this approach, which may be described as “dumb network slicing”, will not be the basis for the new business models that mobile operators desperately need, it is will play an essential part in the End-to-End framework that is envisioned for 5G.
The Application-Aware Network:
A new revenue model for delivering
bandwidth on-demand leveraging
Dynamic Network Slicing.