Talking to a random sample of vendors at Mobile World Congress this past spring, it was tough to get a definitive sense of what the 5G mobile core would look like. There were plenty of ideas and visions, but the only clear point of agreement was that it was too early to know for sure. Well, that and the fact that the 5G mobile core would include mobile edge computing (MEC) and network slicing. Yep, everyone seemed to agree on that.
The fact that many definitions of network slicing include the term ‘network slice’ might seem tautological (or redundant in non-SAT speak). Chalk that up to the concept of a ‘slice’ being particularly apropos for describing dynamically created (sliced) virtual networks in support of specific service requirements. Think stitching together diverse network resources – virtual or physical; RAN, core, or transport; storage or compute – in order to deliver on the demands of network operator or third-party delivered services. Per a recent 5G Americas whitepaper, “Network slicing, in its simplest description, is the ability to tailor a set of functions to optimize use of the network for each mobile device. All of the functionality needed, but only the functionality needed, is assembled in a way that optimizes that device’s ability to find the correct network, access the network efficiently and securely, and be attached to the core network with the set of functionality needed by that device.”
Source: NGMN
Huawei’s network slicing demo with Deutsche Telecom from Mobile World Congress this year helps to further explain what it is. Per the vendor, the slicing implementation demonstrated the notion of “one physical network, multiple industry services,” with a slice manager covering the RAN and core supporting “slice management and slice monitoring.”
It’s a neat concept. And, while some media outlets described Ericsson’s slicing vision as “ambitious” and “far out” when they talked it up three years ago, the concept has gotten a lot more concrete and more broadly accepted. Along with the NGMN, the 3GPP – for example – has been active in defining slicing use cases and requirements which spell out what’s expected from the technology. Requirements like:
- The composition of slices from independent sets of network functions and parameters.
- Dynamic slice creation to address diverse service demands.
- Operation of slices in parallel without negative impacts to data communications.
- Application of security parameters on a slice by slice basis.
- Authorization of third parties to create and manage slices via APIs.
- End-to-end (RAN and core) resource management for each slice.
At this point, the importance of network slicing to 5G should be patently obvious; where 5G comprises a myriad of diverse applications, network slicing is positioned as critical for delivering the specific network capabilities those applications need.
Returning to the fundamental 5G use cases, the role of network slicing is straightforward.
— Massive IoT. Our previous posts pointed to 5G IoT requirements which include massive cost reduction, network scalability and edge analytics. Setting up a network slice that touches MEC-based analytics and security assets while leveraging only the needed mobile core resources (e.g., no mobility support) will be key to delivering on those requirements.
— Critical Communications. Key to supporting critical communications use cases will be supporting reduced latency and traffic prioritization – through the core and RAN. While non-critical traffic might reasonably be supported on distant core assets, slices will be important for tying critical communications traffic to assets in close proximity along with the orchestration of RAN assets to support priority traffic (think appropriate spectrum, location of baseband processing, hetnet layer, SON prioritization, etc.).
— Enhanced Mobile Broadband. Enhanced mobile broadband is about more than ‘fat’ (high-bandwidth) pipes. It’s about delivering rich mobile experiences. Again, slicing will be key to the extent that it can stitch together the best (scalable) core assets and non-congested RAN assets and any content cached at the network edge in order to ensure that latency or network bottlenecks don’t impede the experience.
As much as network slicing may be critical to 5G use cases, the real reasons it is getting so much 5G attention may have nothing to do with specific services or use cases. It comes down to costs and the larger vision of 5G as a service enabler.
On the service enablement front, we need to think about network slices as extensible beyond traditional mobile service providers. Slices specified by third parties would potentially open 5G networks to a new universe of service provisioning and applications meeting the needs of diverse customers in ways that traditional mobile service providers cannot. On the cost front, it’s fair to ask how operators will build a next-generation network while keeping service costs reasonable. Where network slicing can ensure that only the resources needed for a given application are touched, those assets can be scaled independently, keeping costs in check. In a world of monolithic network functions hosted on proprietary platforms, this type of flexibility might not be possible. But, as virtualization becomes the norm, networks become ripe for slicing. And as we move to co-called ‘cloud native’ architectures based on microservices, the applicability to slicing only grows. There’s no doubt, then, that we can expect to see network slicing make its way into more trade show demos and 5G proofs of concept in the near term.
With LTE networks enjoying upgrades that promise gigabit performance and scalable IoT support, it’s often asked what 5G brings that 4G doesn’t. Thinking beyond advanced 5G technologies and incredible new speeds to network innovations enabling a new class of 5G service providers and ‘open’ service instantiation is one answer. And perhaps that’s the reason everyone agrees that network slicing is a 5G core requirement.