A whole lot it turns out when Azhar Sayeed, Chief Technologist, Red Hat, asked Shamik Mishra from Capgemini Engineering for his insights.
AS: Shamik Mishra is Vice President and CTO, Connectivity & Offer Leader, 5G and Edge. As an integrator, software stack supplier, and software developer, what does 5G mean to you?
SM: It’s a major opportunity for us. We work in more than 11 industry sectors, from manufacturing to telecoms, automotive and transportation. Nearly every sector wants to deploy 5G in some way. It’s an opportunity to take our connectivity knowledge and capabilities into these sectors. Anticipating that, we did research through the CapGemini Research Institute, interviewing more than 1,000 executives from industrial organizations to understand what 5G means for them.
Over 60% of respondents said they are early adopters of 5G: some are in the R&D phase, some are doing proofs of concepts. It’s extremely encouraging.
5G adoption is picking up fast. We were hampered by the pandemic a bit but things are back on track.
AS: What are the main obstacles to industries adopting 5G?
SM: There are issues to navigate for enterprise networks as most have not used 5G. How they will they integrate IT infrastructure with OT [operational technology] is an issue that varies across different use cases.
If an organisation has invested in devices that have nothing to do with 5G, do they trash them or come up with new kinds of gateway architectures to connect non-5G devices to 5G? There aren’t any gateway architectures or aggregator architectures that can do that yet.
Also, the same use case with an architecture that works in automotive will not work in a manufacturing. The languages are different, the platforms are different. How you integrate the larger UI [user interface] or the OT ecosystem is extremely different.
There are fundamental challenges in designing 5G use cases, which cannot just be a connectivity stack; they have to be an end-to end-solutions, and solve a problem.
Most enterprises are using cloud native technologies to build applications. How can those concepts be brought into the 5G ecosystem?
AS: So there are barriers to adoption, is there a tipping point in terms of deciding to deploy in the RAN first, or the a core? Where do I start if I'm a network provider?
SM: In the wireless market we’ve seen innovation from outside telecoms. When we used 3G with Nokia Symbian phones, we didn't know how to consume the internet. Then Apple came [with the iPhone] and the Android platform came along with apps, then suddenly we knew how.
4G brought more reliable internet so then we could do mobile commerce. In India, for example, nearly everybody uses uniform payment interfaces – you can live without cash. And you can hail a taxi almost anywhere in the world now too because of mobile commerce.
For 5G to succeed, operators are perhaps too static for consumer industries so they must go for the B2B market. With 3G and 4G, we saw the network came first, then innovation followed.
So the 5G network has to come first, and the radio network before the core so that industries will invest in a couple of labs to get started, and that innovation engine can be energised.
AS: Let’s dig a bit deeper into this RAN and the core conversation. Does cloud plays a role in this and, if so, what is that role? What is the impact of cloud and virtualization on the deployment of RAN and core deployment?
SM: Since 2013, we’ve been trying to virtualize the network and we have seen a lot of modernization efforts going into NFV [network functions virtualisation], but for a mobile network operator, that was only targeting say 30% of the network – the RAN is 70% of their networks and has not been virtualized by most operators.
The benefits of virtualization – lower costs of ownership or better innovation models – completely missed the RAN ecosystem. If you want to benefit from the overall economies of scale of cloud, RAN cannot be left out. This will address the elephant in the room – how to reduce RAN costs?
Secondly, it will open up more innovation on the RAN. In mobile networking, innovation has centred around spectrum and the radio network itself. When building use cases on top of the radio network, how can we improve the quality of experience? How can we create better spectral efficiencies? How can we leverage indoor connectivity and improve it?
RAN virtualization is the most important element here, and O-RAN’s RAN intelligent controller (RIC) is driving a lot of such innovation.
The other part of cloud is the edge, which to me is little clouds everywhere, all over the geographies, and you can run any kind of application on top of it. It’s just extending the benefits we already derived from virtualisation in the core to the cloud, but there’s a significant difference in edge: the applications are going to be cloud native, built by developers who don't understand 5G.
Adopting cloud in its true form – as a cloud-native application developer experiences it – is imperative if operators want to make money from edge.
AS: Open RAN and the O-RAN Alliance is working on standardisation of open interfaces, etc. Do you believe O-RAN is really open, or is there more to be done?
SM: It’s work in progress. Fundamentally O-RAN is working to achieve the flexibility to manage the network better. To do that, we need three things. First, most of the radio network must run on cloud native architecture so it's scalable and Agile, and you can change software easily and leverage industry-tested models like DevOps on top of a radio network’s workload.
The second part is that to disaggregate hardware and software, you need open interfaces, so you have to define and standardise APIs. O-RAN is doing a great job, but it’s work in progress.
Third is how to make the best use of the new open ecosystem? You need better orchestrators for RAN controllers and to collect data from the RAN itself, then leverage the data to build new use cases. That's the whole idea of xApps in building new applications on the RAN itself.
If the operators commit say nearly 80% of the network to Open RAN over the next five to 10 years, it will become a virtualized and open RAN, and operators will benefit. If it is just a proof of concept or one or 10 or 100 sites, the cost benefit will not be realised, but the innovation could still help.
AS: Talking of tipping points, you’re saying if we push for economies of scale, the solution will be better?
SM: Correct. The hyperscalers became such big players because there was innovation, then the platform for innovation provided economies of scale. You cannot do one thing at a time. Both things have to happen.
I would be very surprised if virtual RAN doesn’t become 70% of the overall radio network deployment five years from now. If that doesn’t happen, we technologists have really goofed up something.
AS: There are a lot of conversations around private 5G networks and hyperscalers competing against telcos. How do hypercalers fit into this 5G mix and how are telcos are looking at that option?
SM: Hyperscalers will be there in some form or other in 5G networks. In public networks the core is well suited to be [hosted] in the hyperscalers as long as the regulatory aspects are taken care of. For edge compute, the ecosystem of applications developers globally use the cloud native platforms of hyperscalers. Whether they use Red Hat OpenShift on top of it is a different matter, but the default sandbox for an application developer is on cloud.
For a private 5G network, you still need hyperscalers at least for applications. Going deeper into the network and towards the radio network. hyperscalers have a strong role to play there too, but we are still exploring that space. It depends a lot on how much the hardware and software disaggregation is possible…and how much the vendor ecosystem accelerates the disaggregation.
Secondly, what kind of new architectures hyperscalers can bring to the table for the O-RAN OpenRAN ecosystem – or will they propose something drastically different that we haven't thought through yet?
AS: There are projects like XGVela and Magma for Open5GCore and you have played a role huge role in developing a core stack yourself, partnering with a hyperscaler. What is open source’s role and can it accelerate the development of 5G?
SM: Open source has been a mixed bag for telecoms so far. The cloud native ecosystem and open source have to be mainstream for the simple reason that Kubernetes, the fundamental software, is open source.
To reduce cost and/or make more money out of 5G, we must empower the application developer [and] figure out a way to build use cases. The average application developer is not associated with operators, so operators need to them and what other way is there than open source?
Also, open source enables innovation but if it is not adopted at a scale of, say, 70% of the operators, people will lose interest. Operators have a major role to play and they have to commit to it.
Hear more from industry-leading experts at Red Hat's Open5G2021 Virtual Telecom Event and find out more about Capgemini’s research on the 5G Industrial Revolution and Capgemini Engineering.