Here's how to make Network Slicing work efficiently
You’d be forgiven for being a little bored with the subject of network slicing. The industry has been talking about it since the beginning of the 5G standards process, but until recently not much has happened. That’s because 5G core network deployments lagged behind 5G radio rollouts, and handsets couldn’t support slicing at all. But now handsets are starting to add that capability: Android has been expanding its support since Android 12, and there have been indications suggesting that Apple is adding slice support with iOS 17. Meanwhile, telcos with 5G SA cores are starting to plan services underpinned by network slices. But what other automated slicing orchestration and assurance components need to be added to the network, hopefully without breaking the bank?
Slicing Orchestration Must Work Well with Others
The first thing to know about slicing orchestration is that no telco is going to replace pieces of its network to enable it. Most are either still deploying 5G across their coverage area or are still in the planning stage, so any new slicing enabler will have to work with what’s already there. That means that any slice orchestration component will have to work with different network architectures, different service strategies and priorities, and different business models. It must handle all the “multis:” multi-vendor, multi-domain, multi-technology, multi-edge, and multi-cloud.
All this, of course, while it also accommodates the different ways that operators roll out their slicing capabilities. Most operators will start with eMBB slices that enable fixed wireless and enterprise services, but even within that common direction, the number of slice definitions and how they are deployed will vary widely from operator to operator.
Slice Lifecycle Management is Too Complicated for Humans - Automate It
Network slices are complex, containing dozens of parameters and potentially serving many thousands of devices over hundreds of instances. Creating a new slice definition requires compliance with myriad specifications as well as familiarity with all the network resources that will support that slice. It also requires the technical ability to translate an SLA into a slice definition that meets its business requirements. In other words, it requires intent-based design that must then be passed on to the operational side.
And that’s the easy part. At least, the fact that it doesn’t have to happen in real time makes it relatively easy. It can even be done manually, although most operators will be much better served by using some sort of low- or no-code slice creation engine, and/or predefined templates that can be configured to the telco’s needs.
Operations, of course, must be done in real time. The lifecycle starts with deploying the slice across the network; that process has so many variables and touches so many systems that manual provisioning is bound to produce an unacceptable number of errors. Avoiding fat-fingered mistakes already argues for end-to-end automation; but the need for automation is unescapable once you start to consider slice assurance and orchestration. Slice assurance must monitor the performance of the slice in real time and adjust network resources to keep its performance within spec. It must therefore continuously poll resource availability to predict possible service degradation and aid network planning. And when a given resource fails, the slice orchestrator must select the next best available resource to keep the slice performing according to its business intent.
Slice Orchestration Should Serve the Commercial Side as Well
One of the new demands that network orchestration must address is cost-effectiveness. Our research at GlobalData indicates that the industry is waking up to cost as an orchestration variable. This is a sign of maturity: everyone knows that energy savings is a priority, for example, but slices must also optimize the cost of resources to meet a given slice’s performance. When one of the slice’s network resources fails, therefore, the orchestrator may not only need to choose the next best resource from a performance standpoint, but also from a cost standpoint. It may have to weigh SLAs and commercial rules to decide which slices are degraded first, and in cases like network sharing, it may have to rebalance traffic to spread the pain equally across all the stakeholders in a network.
Nokia’s RAN Slice Controller and Core Slice Controller are self-standing, open standards-based modules that can be deployed independently in a multivendor network. Each is based on Nokia Digital Operations software to take directions from the main orchestrator and apply them to their designated domain. They handle the complete create/modify/delete slice lifecycle and assure slice performance in that domain. These components, designed for seamless integration into 5G networks and support for automated end-to-end slice operations, can benefit operators and generate slice revenues more immediately.
Final Thoughts
GlobalData recommends that all mobile telcos think of slicing as a tool to start generating revenue soon, not as something that needs to wait for the next revenue refresh. Yes, slice orchestration and assurance are complex, but technology is increasingly available to automate away most of that complexity. Telcos should choose solutions that enable intent-based networking and that handle commercial as well as technical logic – that is the way to true network transformation.
To learn more please visit Nokia’s website:
