logo_header
  • Topics
  • Research & Analysis
  • Features & Opinion
  • Webinars & Podcasts
  • Videos
  • Dtw

How Orange’s CTO is driving network transformation

Laurent Leboucher, Orange Group’s CTO and Senior Vice President Orange Innovation Networks, discusses the evolution of the operator’s networks, including similarities between network softwarization and IT system transformation, and shares insight into Open RAN and 5G Standalone plans.

Joanne TaaffeJoanne Taaffe
27 Jan 2022
How Orange’s CTO is driving network transformation

How Orange’s CTO is driving network transformation

Laurent Leboucher has a huge task on his hands. As Orange Group’s CTO and Senior Vice President Orange Innovation Networks he is responsible for steering the evolution of the operator’s networks, including their software transformation.

But it’s not the first time that he’s been involved in largescale software-based systems transformation. Prior to stepping up to the role of CTO in September 2021, Leboucher was Vice President Digital IT, Customer Relation Solutions and Global Architecture, Orange Labs Services, where he played an instrumental role in transforming Orange’s IT systems. “On the network side we are at the beginning of a significant transformation of our industrial model. The software transformation … of the network is something that started some years ago on the IT side. The network is really our core factory within the company and it’s quite exciting because I see that on the network side we are leveraging the same kind of techniques” as on the IT side.

His aim is to take Orange far beyond the simple reduction of network infrastructure costs. “When we talk about cloudification we think of it as a way to optimize capex on the hardware infrastructure. But I don't think that is the most important benefit,” explains Leboucher. “We want to do it because it represents a real change in the core operating model and will allow us to deliver [services] in a way that is much more continuous, automated and resilient.” As with its IT systems, Orange wants to put in place a horizontal, continuous integration/continuous (CI/CD) approach to network delivery so that it can greatly accelerate existing integration processes, all while automating network operations. “For that, of course, we need a telco cloud infrastructure and we need the network workloads to be containerized,” explains Leboucher. “And we need to apply the right patterns of cloud native to benefit from automated resilience and automated scalability and really be able to split the workloads into containerized micro services. And we want to apply policies to automatically activate/deactivate the containers according to certain criteria.”

Orange is testing how its vision will work in practice on Pikeo, an experimental, fully cloudified 5G standalone (SA) network with end-to-end automation, which Orange launched at its lab in Lannion, France, last July with technology partners. Orange and its technology partners are developing Pikeo’s network functions as microservices deployed in cloud infrastructure in containers orchestrated by Kubernetes. The access network is based on Open RAN principles and technologies. “Pikeo is a very important strategic project for us. It's much more than experimenting with a 5G network. It is really about trying out a new way to operate,” says Leboucher. Since launch, Orange has been able to completely automate Pikeo’s 5G network deployment, which Leboucher considers to be a very important milestone. Not only are deployments automated in a matter of a few minutes, the Pikeo network also demonstrates how automation can improve resilience. “If there is a bug and we need to do rollback, it is much easier and we are also able to recover much faster,” explains Leboucher. Machine learning and AI techniques, meanwhile, will help Orange “automate much more of what we do today in terms of network supervision, correlation, incident management, problem solving, but also how we can optimize capex on the network side, and how we can be much more energy efficient”, says Leboucher.

5G SA and Open RAN

When it comes to upgrading Orange’s existing commercial networks, 5G SA networks are among Leboucher’s priorities for 2022. Building a 5G SA, for example, will allow Orange to dynamically slice its network and define logical networks for customers’ specific applications and quality of service requirements, explains Leboucher. Orange is an Open RAN advocate. In November it inaugurated an Open RAN Integration Center, where companies from the Open RAN ecosystem can test and validate their products and services. The operator also joined Deutsche Telekom, Telecom Italia, Telefonica and Vodafone in calling for the EU to prioritize Open RAN.

Nonetheless, Leboucher is realistic about how quickly it can install Open RAN in existing networks. We are working to create that Open RAN ecosystem and make sure that starting in 2023 we'll be able to deploy Open RAN technologies in the network beginning with rural and indoor deployments,” he says. “It's not something that we can deploy to fully replace existing RAN in which we have already invested recently, so it will be step by step.” The RAN integration controller [RIC] will play an important role as Orange’s network infrastructure becomes more cloud-based, horizontal and distributed. “The RIC is key because it's also a way to automate configuration of the network at a scale which is just impossible to think about if we had to do that manually,” he says. Another functionality of 5G that interests Leboucher is massive MIMO, which will help optimize network efficiency by using intelligent beam forming to target users very precisely.

Working with hyperscalers

Increasing virtualization and cloudification is leading communications service providers (CSPs) the world over, including Orange, to partner with hyperscalers. But Orange wants hyperscalers to remain one option among several, says Leboucher, adding that in a number of countries, moving core networks to public and hybrid cloud infrastructures would probably not be allowed for sovereignty reasons. At the same time, Leboucher recognizes that hyperscalers are making rapid advances in understanding telco workloads, while developing expertise in areas such as private networks. But he adds: “I think it's important that there is not only one choice [and] that operators and the network vendors together build and operate alternatives to hyperscalers in the form of telco cloud solutions that can be completely standardized with opensource components,” says Leboucher. “We want to use and leverage [hyperscalers’] infrastructure as one option, but not as the only option. It’s important for us to keep the operation of the network on our site because it’s our core business.”

In the meantime, Orange is very active in virtualizing and containerizing network functions as part of its network cloudification strategy. It is starting with core functions, and is also working on network functions that are adjacent to IT, such as application services. Orange is also addressing workloads such as firewalls, VPNs, CDNs, “and then step by step we start to include also the core function of the network,” says Leboucher. “The trickiest part is the access and especially the RAN,” says Leboucher. “But with Open RAN we have started to use V-RAN capabilities and started to containerize and virtualize control-like functions.” Leboucher is also exploring how network transformation – and the new services, uses and partnerships that future networks will support – will reshape how operations work. “Without automation it will be impossible with the number of people that we have today,” in part because traffic is increasing at an exponential rate, explains Leboucher. “Watching screens all day will be replaced by smarter operations where people receive tickets that correlate to real issues and they can then drill down and find the root cause of the problem, or automatically react.” And he believes new applications and B2B critical services are likely to require very careful SLA management. All of which will mean retraining. “For that you need new skills. It is not something that you can invent just in a lab. You need to be very close to operations and understand what can be automated, how you can close the loop.”