Inside Vodafone’s new global data platform, powered by Google Cloud

By Derek du Preez

May 10, 2021

Audio version

7 min reading

Telco giant Vodafone operates in 22 countries, has partner networks in a further 48 countries, and brings in over €44 billion annually. And as is easy to imagine, in 2021 and beyond, data is central to the its business success - both in terms of improving operations and driving new business models for customers. 

Vodafone has been undergoing a multi-year transformation plan that is seeing it move all of its systems and data to Google Cloud, with the aim of creating an integrated data platform called Nucleus. The aim is to create a single data model that has access to all of the company's data, whereby Vodafone engineers can more easily create and deploy AI models in the cloud. 

This is a complicated task when you're considering 70 petabytes of data and hundreds of internal company systems, all of which needs to be extracted and loaded into the Nucleus platform. As part of Nucleus, a new system called Dynamo is being created to drive data throughout Vodafone to enable it to more quickly offer its customers new, personalized products and services across multiple markets

The telco is now also in the process of migrating its SAP instance to Google Cloud, which will allow it to open up and consolidate its transaction data into this single view. 

We got the chance to speak with Ignacio Garcia, CIO of Corporate Information Systems at Vodafone, about the company's plans and how it's thinking about its future data needs and requirements. 

Garcia explains that up until 2015 Vodafone had carried out a three year transition to put all of the company's processes onto a single SAP instance, which he says was challenging due to the diverse nature of the telco, having grown through acquisition and the development of multiple systems. He says: 

We created shared service centres to support that in a unified way. We standardized our processes and we have a massive backend SAP system. That was finished in around 2015. Then we created a new strategic plan, asking - what is next? How can we bring more value to our customers and to improve operations? 

As part of that we decided to move some components to SaaS - the Aribas, the SuccessFactors - and keep the central instance for finance and transactions. And then we did a migration of that as a brownfield project into HANA. That was a massive implementation, very complex and challenging. But we succeeded and we had our HANA instance on premise. 

Need for more flexibility 

However, despite this success, Vodafone still felt that it had limited agility when it came to executing new projects, which was proving expensive. This was because it needed to replicate that massive system that it had on premise to do any testing and development of new environments. Garcia says: 

There were constant changes, but it was very slow and costly. The second thing is, for a telco, the data you have in SAP will be about 10% of your overall data. In a telco, your billing system is in a separate system, your CRM is sitting in a different system, then you have all the network data. So we created a super ambitious plan to unify and harmonize the data model of all of these systems, consolidate them and create one tooling that is our innovation machine, that can operate at speed and at scale. It's called Nucleus. 

Nucleus sits in Google Cloud and has three core components. Firstly, there's the Neuron platform that houses all the data. Dynamo is responsible for overseeing the data flows and migrations into Neuron. And then thirdly, Vodafone is creating a common data model too. Moving SAP to Google Cloud will allow Vodafone to consolidate all of its company data in one place and execute on new models more quickly. Garcia explains: 

This is important because by moving SAP to the cloud, we are going to accelerate innovation, as complex data movement will be in the same place. 

To copy all the data [every time], it's not sustainable. You do not scale, you do not generate business value. Moving SAP to Google will give us the ability to inject all these engines natively. 

SAP is typically a closed system, and anything you want to do is slow and expensive. It's not because of SAP, it's just because SAP is a system of record. And while it is very good at processing, it's not very good for getting analytics and AI on top of that. With Google, the environment that we have is completely different. We have all of the tooling to create AI.

The bigger picture

However, the ongoing SAP migration is just one piece of the broader Vodafone data objective - to have a single view of company data, to enable speed, flexibility and reduce costs. Garcia explains:

So we have completed the Neuron programme that we started two years ago, migrating all of the big data platforms that we had in Cloudera into Google. We are talking 70 petabytes of data into Google Cloud, on one schema. We have one way to move data around, we have one way to do the anonymization, we have one way to do the encryption, we have one way to do the tagging, and we have one data model. 

What is next after that? Now we have the second big programme, which is Nucleus, which involves moving every single data warehouse that we have in the company into Google Cloud. I'm talking more than 150 systems, that is the magnitude. Why? My ambition is that we will only have two copies of the data in the future - one in the source system and one in Google Cloud. That is not the case today. 

What that means is that we will have one single version of the truth. At the moment you have to reconcile between the different data warehouses. Here we are talking about one copy of the data and it is available to do the magic. So far we have Portugal live, we are doing Hungary and then we will do the UK. That will be a three year programme. 

Garcia says that the success metrics for this will be twofold. Firstly is to reduce costs, or as he describes it, Vodafone's "economical target". And secondly, is to improve speed to market. Vodafone knows the current time it takes to bring a project or product to market, but with the new data model, it has ambitions to improve on this. 

The project with Google Cloud is also impacting how its technology function operates internally, with plans to move to one central team and to shift the focus from maintenance of legacy, towards building AI models that add value. Garcia explains: 

The good news is that we are in the middle of another transformation and this is the accelerator. The company is moving into one technology team. In the past we had fragmentation, with local markets and Group. Since the start of April we have an ambition to move to one technology team and that technology team has domains. 

So you will have CIOs in the local markets, making sure that customer requirements are being met, and that intimacy with the local requirements is there. But you will have domain to do things at scale, to do things across the company. The way I envision us operating is that we will have far fewer systems, as well as deploy once and deploy many, because we are standardizing even the data. 

And that means we will have higher skilled engineers managing the platform and higher skilled engineers doing AI. We will repurpose from a lot of people managing a lot of legacy, to the same number of people building AI models on top and actually getting a lot of value for the company. This will require retraining, getting the right skills, and we've had to create a competence center in Madrid. And even with the help of Google to find the right resources.