Network-as-code: Can it let the 5G genie out of the bottle?
- Network-as-code is being promoted as a saviour for 5G as enthusiasm for network slicing has dimmed
- Will telcos open their networks to collaborators and expose their network capabilities?
- And if they do, will it unleash a 5G genie to grant the world’s telcos their revenue growth wishes?
Even after all the billions spent on 5G network infrastructure by communication service providers (CSPs) and the accompanying over-exuberant promotion of 5G’s capabilities (remember the race to 5G?) the revenue payback for fifth-generation cellular has not yet matched telco expectations.
Until recently, any reservations were usually batted away with: “just wait until network slicing really becomes available, then 5G will change everything.” But as time has ticked on, the industry has become less sure that slicing, as originally envisaged, is the sure-fired path to innovation and high-value corporate network services. Something new is required. That something may involve the application of network-as-code, which in many ways builds on the key building blocks of network slicing.
Network-as-code starts with the network functions and cloud native software-driven concepts found in network slicing and adds a couple of crucial missing ingredients. The first is network openness, so that the distributed service chains it engineers might ultimately work across the different 5G networks and even across different network types, such as fibre, as well as other fixed access and transport networks. That, of course, is something the CSP community is working on right now – see The time is right for network API services, says Telefónica’s Garcia.
The second ingredient is developer engagement, which recognises that services need not be packaged or created exclusively by the network operator or vendor, but may also be configured by the software developer to best match the needs of particular applications. After all, it’s one thing to proclaim an ability for users to engineer their own services on a next-generation infrastructure via an application programmable interface (API), but another to engage with developers to build interest and furnish the necessary tools to weave those advanced service capabilities into their applications to return optimum performance, balancing speed, reliability and cost, along with enhanced sustainability and network security. (For more on this, see Data and application mobility is the key to the telcos’ future.)
So it could be that the telecoms industry, in formulating standards for 5G, have been looking through the wrong end of the technology telescope and evaluating technical approaches from a service provider, rather than a corporate user, standpoint.
Network-as-code comes with an expectation that cloud-native dynamism will eventually enable network functions to be swapped in and out as required. So rather than baking functions into a network slice to match specific vertical industry needs and then attempting to sell the resulting product to corporates – an approach which never seemed the most sensible, since application use within each vertical probably differs markedly from company to company – network-as-code envisages that corporate software developers will be able to apply (or have an artificially intelligent agent apply) a chain of network functions to extract the best performance for their specific networked application.
According to Nokia, one of the concept’s big supporters, network-as-code will involve an “ordered sequence of service functions,” including virtualised elements, such as firewalls and load balancers. Dynamic placement and selection of these functions means traffic can be routed through an appropriate function chain to maximise efficiency, performance, power saving, and so on. So chains will be specified for individual applications, with machine learning and AI tools applied to continually improve performance and adjust to updates and revisions to the application over time.
Another important draw for developers is that network-as-code (if all goes according to plan) will become standard enough to be applied to different underlying networks, including fixed fibre – clearly this is a must with fibre access networks on a charge and the apparently unstoppable rise of AI applications such as Chat GPT, which seem set to gobble astounding compute and power resources.
Nokia’s definition of network-as-code is: “Extreme simplification of network capabilities to make them part of applications,” which neatly sums up the intent and understandably skips neatly over the morass of technical detail that must be negotiated.
It claims network-as-code will “enable the building of a new business and technology framework, elevating the concept of distributed service chains to the mainstream in telecoms… this facilitates the inclusion of developers and the vertical technology partner ecosystem that is foundational for value creation. It brings new business value creation to 5G cloud-native adoption, openness, and the programmability momentum in today’s telecommunications domain.”
To get there, though, key issues must be hammered out and obvious obstacles overcome.
According to Caroline Chappell, research director at Analysys Mason, who discussed this topic during a recent TelecomTV event, one part of the problem is that there are so many network use cases and moving parts, all running across different clouds, that the industry faces a real fragmentation problem when it comes to opening up network capabilities via APIs.
There are “APIs everywhere”, she says, but what’s missing is a coherent, consistent environment so that developers have the confidence to start using them, a point reinforced by Shkumbin Hamiti head of network monetisation for Nokia’s cloud and network services platform business, who shared the panel with Chappell. Users, he said, need an easy way to connect their applications and processes to consume the available connectivity, be it private networks or telco slices, without having to be 3GPP technology experts. Network-as-code is therefore, clearly a step in the right direction.
According to Chappell: “There’s a long way to go, but if you think about the standalone 5G core, that is a cloud native software-based network. It’s software; it talks API natively and is no different from an IT application that talks API natively.
“The eventual ability to treat network-as-code in the same way as you can treat cloud-infrastructure-as-code, and then being able to put them both in the same application pipeline, is an amazing vision,” said Chappell. “There are many steps and challenges ahead but that’s what we see as the network-as-code end goal.”
For a deeper dive into these complex issues, view Unlock your network’s value with Network as Code, a panel discussion with a question and answer session.