The softwarised network and how to get there

Extracting the true value of the softwarised and cloud-native network. Panel debate at MWC24 with Telefonica, Intel and Mavenir.

Meet the speakers:

  • Udayan Mukherjee, Intel Senior Fellow, Network and Edge Group, Intel
  • Cayetano Carbajo, Director for Core, Transport and Service Platforms, GCTIO, Telefonica
  • Bejoy Pankajakshan, EVP, Chief Technology and Strategy Office, Mavenir


What do we mean by the softwarised network?

It is so we have the advantage of automation, for instance that we can do automated testing and deployment – CI/CD – so we can change the software without having to enter maintenance windows. And everything can be orchestrated. It makes the technology more easy to operate and more agile. We want to take advantage of all the cloud technologies – automation, reliability, agility. We have to bring these advantages to the network.

What’s driving softwarisation can be multiple reasons. One is observability and being able to collect data without custom tooling. Being able to re-use the hardware for different purposes and generate the benefits of scaling software up and down. That will ultimately get you to monetisation, leveraging data for new use cases. But you cannot do that if you are not softwarised.

The whole premise starts with disaggregation of hardware and software. From there it’s about increasing utilisation using cloud technologies. Now you have a platform which you can build many use cases on top of. You can do this because it is a softwarised network.

Why are we doing this?

When you have a softwarised network there are more possibilities for exposing network functionality. Network APIs are a key element of our plan for monetising. Telefonica has been very vocal about exposing APIs in common with other operators. Orchestrating the network and making a flexible network that can have an answer to many different requests in a dynamic way is also a key element that will enable monetisaiton. I’m referring to slicing. The fact we can dedicate a slice for a given service can bring us a very powerful tool for monetising.

What we have today is manual slicing and we are already monetising that – and have revenue coming from that today. The plan is having more automation, more dynamic slicing at the end of this year. We believe by the end of 2025 we can have closed loop automation, Level 4 TM Forum automation. We are working for that.

To truly get to dynamic slicing, on-demand slicing is about tying together network analytics with charging analytics. Bringing this together will enable these use cases, knowing how the user is behaving, how the network is and being able to adjust for SLAs. Getting to that requires us to do these network and charging analytics together. That’s a key next step that we see.

What’s difficult and challenging? What needs to improve?

It’s getting there. There is always going to be refinement in APIs, but you will see a lot more use cases coming.

The situation today is the majority of operators are using the CaaS from the vendor in the majority of cases. We don’t have the transversal aproach. The effort and the cost of certifying on so many different platforms is not making it fly. When you count number of demanding network functions on top of transversal CaaS, there are not so many cases in the world. This is a pity because we are losing capabilities for auto-healing and scaling, or reusing capacity for services. We are not there.

The main challenge for softwarisation is changing the culture and operational model. You have to change functions. There are things in this operational model that have to be changed – but it’s both a need and an opportunity.

We are deployed in different stacks with different operators – different infrastructure. From a maturity perspetive it helps us as a vendor make our software stacks that are more mature running across these different infrastructures. We need to build abstraction layers that allow us to hide the complexity of the underlying CaaS/PaaS layer. When we port our application – if we have to reengineer our software every diferent time that is not sustainable. So getting there is a key strategic investment area for us.

What tangible benefits can we point to?

So for instance we have implemented our pipeline for CI/CD, and this has allowed us to do a couple of things. First is automatic testing and onboarding of new CNFs. A benefit. Second and more important it has allowed us to increase our level of quality. It’s automatic so we are not depending on the human element for parameterising and configuring the diferent network nodes. And this is a tangible benefit already.

All of these discussion start with the TCO, realising the operational component of it, opex cost reductions by, for example, using IT resources. Automation itself has brought in a lot of those benefits. There’s still a long way to go but as we have been partnering and deploying for the last five years, there’s been a lot of learning.

Two years from now, where will we be? Let’s not repeat the conversation

Our target is to maximise the benefit of cloud native implementation in the next couple of years. That doesn’t mean that I will have everything on top of SYLVA or a transversal layer. I am not meaning that. What I will do is extract the maximum benefit of softwarisation. I will have microservices, containerised on top of whatever infrastructure and I will do all the automation I need to do.

I hope we are not having this conversation on how ready we are but actually on benefits we are seeing. Operators today are not making tonnes of new revenue in 5G, a lot of it is about getting to lower cost per bit and to do that you have to use tools like AI and automation. If operators don’t make money, vendors don’t make money and that ecosystem doesn’t flourish like the hyperscaler ecosystems have. So we want to get there and the only way to get there is through automation and softwarisation.