What is 5G network slicing?
Network slicing has been touted as one of the key opportunities enabled by 5G. One that poses a unique proposition – the ability to run multiple logical end-to-end networks on common shared infrastructure, with greater flexibility than before. But where are we today in realising the vision for 5G dynamic network slicing?
What is 5G network slicing?
In our report 5G network slicing: How to secure the opportunity, we defined network slicing as “a mechanism to create and dynamically manage logical functionally-discrete end-to-end networks over common physical infrastructure”. In theory, this means operators can offer multiple networks – each tailored to individual customer needs – to be deployed, managed, and retired as required, on the same shared hardware. This allows operators to have better control over their offerings while delivering them in a more cost-effective way, as the need for dedicated physical appliances for individual networks become a thing of the past.
As you can imagine, with vastly different resource needs (compute, storage, or networking demands) and the added complexity of managing different network lifecycles at the same time, dynamic network slicing requires an underlying infrastructure that is not only flexible and highly automated but also offers full end-to-end visibility of the network. As such, only with the much anticipated fully virtualised 5G standalone network will dynamic network slicing become a reality.
How will 5G network slicing work?
In essence, each individual end-to-end network slice has the functionality of a complete network including specific network layer capabilities, operational parameters, and network characteristics. Naturally, they will each have their own resource requirements as well, such as for compute, storage, or networking. Once deployed, it is known as a “network slice instance” where each slice has at least one instance, which defines the behaviour of the slice. Enterprise customers can then be supplied one or more of these instances based on the slices chosen. This is depicted in Figure 1.
As seen in Figure 1, each network slice can have different network requirements depending on the targeted use case, which defines the structure, configuration, and workflows for the network slice instance throughout its lifecycle. For example, in the context of 5G, a customer connecting IoT devices on a mass scale, whether it is a utilities company monitoring its smart meters or a logistics company tracking assets in a warehouse, would be using a slice more optimised for massive machine-type communications (mMTC) whereas an application involving autonomous vehicles would require a slice optimised for more ultra-reliable low latency communications (uRLLC).
Why is 5G network slicing important?
The value network slicing brings to both operators and enterprise customers is significant. For the former, it largely boils down to service agility, flexibility, and customisation which inevitably trickles down to enterprise customers as well, or taking a step further, even enabling a new generation of ecosystem players.
Service agility: We see network slicing as a key part of telcos’ private networking services menu. Instead of having a dedicated private mobile network (4G LTE or 5G), enterprise customers can use one or more network slices suited to their specific connectivity requirements. The concept of slicing can provide logical separation of virtual networks for different services, allowing different teams to manage and implement changes where they see fit, independent of one another. For example, an individual team wanting to launch a new service that is vastly different from current core services, could do so without reengineering the whole network every time – boosting go-to-market time and potentially reducing cost of deployments. This enables operators to deploy new services in a much quicker fashion to meet customer demands.
Flexibility: The ability to tailor network characteristics according to use cases means diverging from the “one size fits all” approach past generations of cellular networks have taken. Ultimately, this means offering solutions that can better meet the specific needs of a broader set of customers – whether in different industries, geographies, or both. As an example, with network slicing, enterprise customers who want tailored network characteristics with better security features than those offered on public clouds can now opt for a network slice rather than a whole private network solution.
Customised solutions: Network slicing can also be a means of ‘trimming’ the otherwise “bloated” nature of past cellular networks. With the “one size fits all” approach, entire networks are filled to include every function and capability to provide for every eventuality and support all potential applications and use cases at once. With network slicing, operators can tailor networks according to an end use case, and effectively ‘trim’ unnecessary functionality and use the savings to add new technologies that better serve customer needs or even offer more competitive pricing models. For example, an IoT sensor-based application would not require functionalities surrounding mobility or voice communications, which bear unnecessary costs but can benefit from technologies that reduce power consumption.
Enabling new business models: In our report, 5G and MVNOs: Slicing up the wholesale market, we find that network slicing has the potential to even disrupt the nature of an operator’s wholesale business relationships. One area of interest is in how network slicing can catalyse a new generation of virtual network operators such as slice network operators (SliceNOs) who utilise tailored slices to fuel even broader offerings. For example, a car manufacturer could obtain one or more slices for telemetry, entertainment and even remote-control for its future vehicle offerings. Another example would be a government slice operator who has multiple tiers of unique encryption for messages/calls depending on need – the theoretical possibilities are endless. The promise of a flexible, scalable, and tailored end-to-end virtual network is especially an attractive option to even mobile virtual network operators (MVNOs) who, with the help of other players within the ecosystem, could be early adopters of 5G – a key business area for operators where MVNOs now make up around 10% of cellular subscribers in many developed markets.
Tying these together, network slicing is central to the 5G opportunity and it is evident that network slicing has a huge potential within the telco ecosystem, if done correctly. However, it will also both enable and require changes in operators’ business and operating models to support and facilitate the dynamic aspect of slicing: that network slices can be “spun-up”, scaled-up, scaled-down and retired when and where needed.
Addressing the perceivable challenges, whether cultural, financial, or technical, will be essential especially as operators seek to navigate through the Coordination Age, where to truly differentiate themselves outside of the traditional connectivity play, they will need to leverage on the network slicing opportunity, especially with its ability to stimulate new ecosystem partnerships and business models, to drive future revenue growth or to differentiate themselves from other players looking to re-invent themselves.
But what’s the catch?
Fundamentally, the complete end-to-end virtualised network (as envisioned in standalone 5G) needed for dynamic network slicing is not yet been reached. However, even with standalone 5G, we can only expect a fully-fledged network slicing deployment in the following 2-3 years thereafter – for more mature countries. Unfortunately, standalone 5G is only one of the main drivers for greater industry adoption and other barriers also exist.
A significant barrier is the operational challenges and subsequent enterprise scepticism of the security of the network slices. Security in this context refers to the susceptibility of the network to both intentional attacks such as malware and unintentional failures including service failure or simple incompetency. We outline three challenges below:
First relates to susceptibility from enterprise slicing management tools and interfaces. Providing enterprises with the ability to manage their own network slice instances through a function known as network slice management (NSM) means direct control over a slice instances’ lifecycle. Impersonation or unauthorised access of the NSM could lead to manipulation of critical network functions and subsequently severe service disruption.
The second is the disparity in levels of security between the devices hosted on a network slice or between the slices themselves. For example, in the future where millions of devices are connected to the different network slices, ‘lower security’ devices such as unmonitored IoT sensors could become an entry point for network access to adjacent slices which may host devices which contain sensitive customer information such as handphones or smart home devices. The same concept applies when comparing different network slices, where a slice meant for mMTC applications may become a conduit for attacks into uRLLC network slices sharing a similar virtual infrastructure.
Third, is the potential interdependence of network slices through the resource-sharing of otherwise finite resources such as storage, compute, or network functions. If one slice overloads another’s capacity, this may override ones’ ability to run normal security protocol, leaving it vulnerable. These overloads can result from mass reboots or even a denial-of-service attack (DDoS) on one network slice. There is a large concern on how the network will manage itself and prioritise in that scenario.
It is fair for enterprises to be concerned about these, as for those who run more mission-critical services with strict SLAs, service failures could mean significant loss in customer confidence or even worse, jeopardising public safety. However, despite what has been outlined, the concern for security stems itself from a more general issue – network slicing itself is still very new and largely unfamiliar to enterprise customers.
This unfamiliarity is warranted as operators themselves are still grappling to define best practices, commercial models for customers to even the management of the services themselves. This is exacerbated by the fact that even between different operators, the level of buy-in and enthusiasm towards network slicing varies. It is then no surprise that enterprise customers themselves are confused on what network slicing has to offer and are prone to the misconceptions of how network slicing can deliver value to them. It then becomes apparent that to truly capture the enterprise opportunity, operators must understand the challenges their customers face and devise solutions for them. But more importantly, they must first work amongst themselves to define what, how and when they will deploy network slicing.
Need for greater customer awareness and understanding
The industry is working tirelessly to define the architecture needed to deliver standalone 5G and subsequently, dynamic network slicing is becoming increasingly possible. With the security concerns, enterprise customers can be rest assured that when done correctly, network slicing can offer a secure network solution with all the intended benefits outlined above.
To put this into context, with network segmentation, network slices can be limited to a subset of devices and users, allowing operators to apply specific security protocols and controls based on purpose. Furthermore, with the right degree of isolation, network slices can be designed-in so that allocation of resources are specific to each slice. These in combination with the right encryption tools and network authentication and authorisation protocols, having multiple networks can offer enterprise customers a more flexible and cost-effective solution that is also resilient.
What next for operators?
To summarise, there is an opportunity for operators to truly differentiate themselves and capture new opportunities with 5G, but it requires a careful assessment of an implementation strategy, product design and ultimately customer engagement. But to do this, operators must first look to themselves to identify what role they intend to play in the world of network slicing.
Despite the fact that fully-fledged dynamic network slicing is still in development, operators should avoid waiting for maturity when 5G networks are ready to start testing. Operators should look to pilot network slicing for internal use first, then look to start piloting simplified implementations of network slicing for customers with existing technology (even if they are static network slices). These can be upgraded to fully-fledged network slicing when the technology is ready. In tandem, customer-facing teams should start working more closely with enterprise customers to build a much stronger understanding of their business problems, use cases and related connectivity requirements, to build greater trust and confidence for when network slicing is available.
Author: Richard Lim is a consultant at STL Partners with a focus on areas such as telco cloud, 5G and edge computing