CxO Insight

A long game: driving patient urgency to reinvent Vodafone’s networks

Scott Petty, CTO, Vodafone Group, describes how many strands and immense effort, woven into a careful strategy, are building a next-gen telco. By Annie Turner, Contributing Editor

Scott Petty has worked for Vodafone, based in the UK, for just over 14 years. Having held a number of positions, he moved from being CTO of the UK operating company in April 2021 to become Chief Digital and Information officer for the group. He retained that role when he was promoted to CTO of Vodafone Group with dual responsibility for Vodafone Technology in January 2023. The Technology unit has a team of more than 30,000 people spread across Europe, Africa and Asia.

Petty is known as a clear, considered speaker, with a refreshing habit of refuting received wisdom with quiet conviction. For instance, the strategic decision to add 7,000 software engineers when many think the primary motive for network automation is to reduce headcount. Petty says, “This an area I’m passionate about. I think most telcos thought about headcount in the wrong way.

“They counted employees at their company; you need to look at your total workforce, which consists of your own people, the people in your offshore centres, then all the people who work for systems and network integrators to deliver services to your organisation.

“When you take that total headcount view, our overall headcount is going down because we are automating more, but our own Vodafone employee headcount is going up…We realised that if we wanted to win in digital, build platforms and do the sorts of automation we talked about, we needed our own software engineers and our own capabilities. We couldn’t rely on outsourcing and third parties to do that effectively.”

Objectives, not numbers

Numbers are only part of the story, Petty points out: “Our employees are linked to Vodafone’s objectives; of course, external service providers deliver the objectives of their company…If you can break that and take more control yourself, you can move faster and end up with much better quality. Over the last couple of years, our operational performance has improved dramatically and the number of incidents is well down.”

Vodafone is a long way towards meeting that target of adding 7,000 software engineers “and we’re ahead of our business case in terms of cost savings for insourcing,” he says. “There has been some [additions] from retraining staff, but a lot of it has been hiring new people, bringing in graduates building our shared services centres across India, Egypt and nearshore centres in Europe.

“Consequently, our headcount has gone up in Technology as we’ve added those capabilities, but our external spend with systems integrators has gone down by more so it’s helped us deliver our cost savings targets as well.”

Being the B/OSS

In a similar vein, what would Petty have done differently at the start, knowing what he knows now? “Been more dictatorial with vendors on what we expect them to support [and] we would have invested a lot more in our data analytics capability, getting our data sources into a Big Data ocean, then getting everyone to write their APIs and calls to that,” he replies.

“It took a long time to break the siloed approach of OSS vendors and automation vendors who want to have their own datasets, their own data pools, and then make their own decisioning based on top of that. You can harmonise the data and where you put that data, and then use them as a decision-making presentation; that is, you get a lot better automation than if you rely on an end-to end-vendor.”

He says, “[OSS is] by far the hardest, due to lots of smaller vendors with lots of monolithic software stacks that aren’t using open data sources and open APIs. It’s been a real challenge to evolve that area and a lot of it we’ve built ourselves.”

Regarding BSS, Petty adds, “We’re writing all our own digital channels, our own apps and websites. We’re not relying on BSS vendors and all their components…They didn’t give us the flexibility and scale we wanted.”

The road to Standalone

Just how much is 5G Standalone (5G SA) an evolution of the other initiatives towards network automation? “It builds on everything we’ve been doing for the last 15 years – virtualising our core networks using software automation,” Petty states. “We’ve been doing software-defined networking (SDN) to control network elements in 4G, then 5G NSA [Non-standalone] and that became the enabler for us to do 5G SA.

“The industry has been at this for a long time and thank goodness we have because we’ve learned a lot. It wasn’t that easy to virtualise a traditional EPC [evolved packet core] and run it in an automated fashion. It took a lot to get that to work effectively, with the same levels of stability and performance. We’ve needed that time over the last years learning about network automation and virtualisation of the core network to get it to perform appropriately.”

Now another big goal that requires a terrific effort looms with the shift to cloud native “because that’s the only way to take advantage of the real benefits of cloud,” he says. “For auto-scaling, massive horizontal scale, you need to be cloud native; you need componentised software to make that happen.

“It’s taken a lot of work for software vendors to rewrite their core nodes – their GGSNs [GPRS support nodes], their HLRs [home location registers] – all those elements to make them cloud native [because] they had monolithic stacks that relied on a close linkage between their software and proprietary silicon to correct performance.

“As you move to more open silicon and started to virtualise, they had to recode a lot of those platforms to make them work effectively. That brought along some new vendors like Affirmed which was bought by Microsoft, so it started to create more competition in the core network.”

The telco world – and timetable

There are still only about 40 5GSA implementations globally. How far are we from scaling and monetising attributes like slicing and low latency? Petty replies, “I think you’ll see it first in mobile private networks leveraging some of those capabilities*. We’ve seen deals like that in ports, factories, mines and so on.

“In the broader context, I think we’ll start to see scaled applications using those environments mid-decade or maybe later in the decade. That’s the telco world. You have to build the infrastructure, build the coverage and enable developers to build new applications and services that sit on top of them.”

Kubernetes expands

As Kubernetes was designed to containerise storage and CPUs, how big is the challenge to extend it to the network? “[It’s needed] a lot of development, a lot of R&D, a lot of work as an ecosystem company to make it work,” he says. “Open RAN is very similar; as we build Open RAN solutions, there is a lot of learning what has to happen in the software layer to take standard industry components then get them to perform and scale at the level the network needs.”

How painful is that integration overhead? “The first issue of all is the data. Machine learning, AI and automation rely on data and sources of data and being able to learn from those datasets. Having a clear strategy for the collation of that data in a structured model that your tools can then work on top of is one of the key elements for creating that interoperability,” Petty explains

He stresses this is “not just the APIs, but giving lots of different tools access to the same data sets so they can do their very specialised task in a way that makes sense. Sometimes the tools need to talk to each other, sometimes they just need to use the same data set to make the decision that they’re performing.”

Petty dismisses the notion of a programmable network as marketing speak rather than an actual aim in telecoms. He asks, “Can customers call APIs on a network today and provision services? Yes. Can we do that ourselves? Yes and we do it all the time.” The web application firewall (WAF) is a good example.

Vodafone’s website and MyVodafone app requires roughly 200 releases a month. WAF’s configuration is integrated into the CI/CD pipeline which means “as we do those 200 releases, the network elements below them are automatically configured, tested and validated to make sure that as we move our code from dev to pre-production to production, that process happens automatically,”

Scott Petty, CTO, Vodafone Group, is speaking at FutureNet World, 3-4 May, London

Petty says.

“That agility is super important as you move to cloud but also in your own data centres’ environments as you move to Agile and try to speed up the network.”

He concludes, “Our goal is automation and enabling agility; giving people the right tool sets and APIs to make configuration calls into the network, without needing to place orders, and do all those kinds of things in the areas that make sense and invest in them.”

* Shortly after this interview, Vodafone unveiled the jewel in its MWC23 crown: a prototype of an Open RAN-compatible 5G network-in-a-box, which, if productised, could have a huge impact on the 5G private network market.


Wake up Neo…

Contributed by Sam Keys-Toyer, Head of Business and Portfolio Development, Managed Services Networks, Ericsson. The... Read more
Chat with our team