Deutsche Telekom boosts IT productivity with platform approach and cloud-native public cloud

Platforms are doing more to increase developers’ productivity at Deutsche Telekom than any single technology deployment, including Generative AI, according to Laurent Donnay, Chief Engineering Officer, Deutsche Telekom. And their principal strength lies in creating common environments for IT and network developers within an increasingly software-driven industry.

“Platforms are really critical for one major reason: Software is now the core of the telco world. Software-defined networks are done and dusted. If you want to both serve your customer but also drive and steer your network going forward, platforms are arising right in the middle to serve the historical IT world and the historical network world,” says Donnay, adding that without platforms “you can't do IT at scale.”

Benefits of platforms

A platform provides developers with a consistent and easy-to-use set of capabilities, including Open APIs, which allows them to spend more time producing high-quality code and less time configuring a cloud landing zone or a pipeline, explains Donnay.

“It really has a very strong impact on all the teams and on how fast we can create our products,” resulting in a time to market for new features and products of under two months, he adds.

A platform also makes it easier for Deutsche Telekom to integrate new tools.

“Whereas adopting a GenAI tool might free up 10% of engineers’ time to do more coding, the figure rises to approximately 20% when it is made available as part of a broader development platform,” explains Donnay. This is because “it's seamless, integrated, so you don't need to change your environment. You just reduce the number of clicks to get a similar outcome.”

Legacy applications, however, can throw a spanner in the works.

“The landscape is still a very diverse mix of things. We have new stacks that are open source based, fully cloud-natively written … and you have kind of an immediate benefit from all the platforms provided by the hyperscaler stack,” explains Donnay. “But we also have in our portfolio a whole set of applications that are older that we will not be retiring in the next two or three years. The challenge is balancing both worlds.”

Nonetheless, Deutsche Telekom has already transferred 30% of its IT development to platforms, which is helping the company to adopt AI and automation at scale. Donnay also believes that with the “right platform approach — mixing monitoring, observability, IT service management — we could really be strong in avoiding incidents, doing predictive analysis with AI that would make us more resilient.”

Not all clouds are equal

Platform adoption goes hand in hand with running systems in the cloud. Today 80% of Deutsche Telekom’s IT systems are hosted on cloud infrastructure, of which a third are cloud-native and hosted by hyperscalers. The rest are a mix of “lift-and –shift" or containerized systems, hosted on premises.

The company’s different experiences in recent years of moving systems to the cloud have demonstrated to Donnay the advantages of public infrastructure and cloud-native applications.

“The real benefits of the cloud are a cost of infrastructure reduced by 30%; a drastic improvement of resilience and stability and time to market ... Also, a lot of the native AI features — you only get that when you run on hyperscalers,” Donnay says.

“If you develop a tool or software to run on AWS tomorrow, you use the Amazon Q infrastructure to code on it, it's going to be really saving time to get a very strong result,” Donnay explains.

In contrast, lift-and-shift failed to deliver the expected cost savings, according to Donnay, and the move to virtualized infrastructure initially impacted mean time to repair. “It is a pretty brutal decoupling. Where teams in the past used to own an application from the UI [user interface] to the database and own the entire physical stack, they always knew pretty fast what was not running and could fix incidents and mean time to repair was strong.”

Deutsche Telekom is therefore embarking on what Donnay calls a second cloud journey “to take everything that is not yet cloud native … and move [it] more and more to a hyperscaler.”

This strategy is made possible by hyperscalers’ roll out of sovereign cloud offerings in Europe, which will allow Deutsche Telekom to continue protecting data in line with EU regulation even when using a private cloud.

Data regulation also informs the company’s platform choice. Whereas software engineers in Europe will focus on a platform’s data privacy protection, developers outside Europe will look for speed and stability. “We’re lucky because we use the same platform [worldwide and it means] we help each other to serve the final end customer,” says Donnay.

Democratizing innovation

As telcos’ use of software and platforms develops so will the way they innovate, believes Donnay: Although he expects to have the same number of developers working in his team in five years, what they do is likely to change.

“What I really like in our industry today is the fact that the cadence – because it is primarily now software driven – is drastically increasing,” he says, adding that “GenAI wave shows that innovation is available to everyone.

“Going forward in the engineering space, we shall spend more and more time understanding the different trends and to think of how to adapt [them]. And that is a very important shift, because I don't think that we can be successful in the future if we have … an R&D team … doing the work nine months ahead and then giving to the others,” Donnay concludes.