If you look at most telecommunications networks today, they’re divided into a kind of vanilla/chocolate/strawberry mix of services where you may have an enterprise flavor, a standard consumer flavor, and a high-usage consumer flavor. When 5G arrives, however, those networks will need to look a lot more like Baskin-Robbins’ 1,300 flavors. What telcos will do to create those flavors is customize the network experience using thousands of different policy-driven network slices with unique SLAs, security features, latency and bandwidth requirements, pricing, and so on. This blog outlines what network slice management solutions for telco operators and what it should look like in 5G.
Benefits of Network Slicing for Telco Operators
The benefit of network slicing is twofold: to create differentiated services that people will pay for (e.g., high-bandwidth virtual reality slices, low-latency mobile healthcare slices) and more effectively manage network traffic and costs (e.g., IoT and online gaming will clearly have different pricing and priorities). The challenge of network slicing is that this is largely unknown territory for telcos, which are accustomed to cautiously rolling out new services and managing just a handful of pricing models (e.g., unlimited plans, per-gigabyte pricing).
The concept of a network slice technically dates back to 3G, when it was called an access point name (APN). In those days, APNs worked like a giant, best-effort bucket: telco operators would pour their water into a single bucket, poke a bunch of holes in the bottom and customers would get more or less water depending on how big a hole in the bucket they had. From that perspective, I guess you could say that APNs “pail” in comparison to today’s network slicing capabilities.
Network slicing gives operators a much more granular level of control over how they allocate their bandwidth, arguably their greatest asset. They can deliver higher SLAs around revenue-generating apps, give lower priority to video streaming that doesn’t generate revenue for them (ahem, Netflix), wrap security policies around enterprise traffic for added privacy, and so on. Actually, a lot of “so ons” – like thousands of them.
This is very different from the way bandwidth is managed in the 4G world. For example, in a 4G network, there might be one big slice for all consumer services. We’ve even seen telco operators adopt this same model for their initial 5G rollouts. But once you get past the trial stage for 5G, the goal is to increase the number of slices on day two, three, etc. until you have network slices for nearly every imaginable scenario: video gamers, casual users, small business e-commerce, healthcare apps, etc.
Network Slice Management Functions
As you might imagine, how telcos will stand up and manage these slices is a topic of much conversation. Industry organizations such as 3GPP, GSM and ONAP have all weighed in on what they believe slice management should look like in 5G. Basically, it boils down to three key network functions (and you might want to have your alphabet-soup decoder ring on hand for this):
- CSMF (Communication Service Management Function), which acts as the user interface for slice management;
- NSMF (Network Slice Management Function), which controls the slice, end to end, across the RAN, transport and core domains (also referred to as subnets);
- NSSMF (Network Slice Subnet Management Function), which applies the NSMF’s lifecycle management commands (e.g., instantiate, scale, terminate, move) within a particular subnet.
The NSSMF is where most of the slice intelligence resides. It takes a command from the NSMF, such as “build a slice,” and activates it by doing all the behind-the-scenes work of function selection, storage, configuration, and communication. And this brings up another important point that industry organizations have focused on: creating API standards that dictate how all this communication and interaction should take place between the NSSMF and other network elements.
Within 3GPP is a group known as SA5 that is working on a reference architecture for network slice management. We believe it’s in the best interest of every NSSMF vendor to follow these standards (it’s something we’ve done from the beginning with our own NSSMF solution for the mobile core subnet). Why? Because a best-of-breed approach to slice management is the best course of action for most telco operators. Yes, there are vendors that offer a soup-to-nuts solution, but in our experience, no two operators have the same slicing requirements. A one-size-fits-all approach to slicing just doesn’t fit that reality.
We view the best-of-breed approach as following one of two courses. There’s the multi-vendor approach, where operators engage with a bunch of different vendors to populate a complete slice management solution: a RAN NSSMF from vendor A, a transport NSSMF from vendor B, a CSMF from vendor C, etc. And then there’s the standards-driven approach, where an operator engages with an organization like the ONAP Project to build their own network slice management solution. This last approach may have the most potential for success, as we’ve seen in some of our own customer projects.
The Impact of Network Slice Management
Network slice management will have a significant impact on how quickly and effectively operators can monetize their 5G investments. Automation, orchestration, and observability also have important roles to play in the number and kinds of slices your network can spin up. The appetite for customized slices is clearly there. The question is: Do you have the right tools in your kitchen to handle the demand? It’s a topic I’ll discuss further in my continuing series of network slicing blogs.