“Without third-party app developers there is no edge proposition.”
So commented Altran’s VP & CTO Connectivity Shamik Mishra. With huge amounts of research – and marketing dollars – being invested in edge computing, this warning stands in stark contrast to optimism in much of the telecoms industry.
Part of CapGemini since April 2020, Altran has long been a partner to the major telecoms vendors and a research organisation working across industries from aerospace to factories and automotive. 6GWorld™ spoke to Mishra in an exclusive interview to find out what lies behind such a prediction.
Building Business Relationships at the Network Edge
To start with, let’s define ‘the edge.’ Jason Hoffman, CEO of MobiledgeX, set out some of the challenges in this during a video interview last year, explaining that in the broadest definition it is everything in between a device and hyperscalers’ cloud datacentres, or “Everything that the network operator owns.”
However, consumer edge networks are set to move away from an exclusively operator-owned infrastructure. Private networking may relieve the operator of some of this opportunity. Indeed, this ABI forecast predicts spending on enterprise private networks will jump from $0.6 billion in 2020 to $10.6 billion by 2025 and almost $65 billion by 2030. If so, then there is clearly a threat that lucrative opportunities at the network edge will be lost to in-house enterprise infrastructure.
However, Mishra was cautiously optimistic, as long as the operators take the right approach: “Once [enterprises] are connected, they have to have somebody at the end of the data pipe to process the data arising from being connected. That means that the applications have to be placed either at the edge or the cloud; most likely the edge depending on the kind of applications we are looking at.”
There has been controversy about the kinds of use cases 5G and edge (singly or in combination) might offer. What kinds of applications might we be looking at, realistically? Mishra was enthusiastic
“We could use event-based architectures to offload computing from devices to the edge. For example, a voice assistant in your home does this in the cloud: When you ask a question, the assistant sends an event to a server on the cloud; the cloud launches a small container, which then tries to find the answer from the internet and sends the answer back. Then that container closes and your voice assistant delivers the answer. That’s a very simple example of how event-based architectures work in the cloud today.
“Imagine the power we could create by moving that architecture from the cloud to the edge. There could be hundreds of thousands or millions of events created by devices in a factory that could get processed at the edge, which saves money on transporting exabytes of data to and from the cloud. A whole different way of creating applications and digital twins would follow. So it’s a massive opportunity in our opinion. It merges the cloud, edge devices, the way we design applications. All of this will change in the industrial context.”
Developers: The Big If
It’s a tempting vision of the future.
“We strongly believe that edge compute for a telecom operator would make sense… if there is a developer experience which makes life easy for developers to build applications,” Mishra enthused.
Fundamentally, Mishra sees a clash of cultures as the driving force behind this very particular requirement. Where telecoms providers have had a standards and interoperability-based model of operation, the environment that applications developers know is cloud-native: “There are no standards, it’s all about the developers. The developers are the No. 1 citizens in a cloud ecosystem. So, naturally, they have a very different way of working.”
However, the history of developer ecosystems fostered by telcos has not been glorious. “None of them worked, simply because the operators’ language and the developers’ language are different.” Mishra explained, “The operator will talk about five nines, regulation, things like that. The developer says, ‘I need to fail fast and move on.’”
Whose Edge Applications?
If the operators aren’t the natural fit for creating an edge application ecosystem, where would the applications come from?
The obvious answer would be the operators themselves. However, Mishra warned from experience that “they won’t make much money out of that because they simply lack the agility to build applications at scale. So if they want to seriously make money out of 5G they will have to find a way to run applications on their infrastructure and attract a developer ecosystem.”
Where do the telcos find those developers, then? The choices are all challenging for one reason or another. “At some point they have to either collaborate with the cloud players; or spin off start-ups who are more friendly to developer ecosystems; or come to us to get a developer ecosystem or applications built for their customers. The final model would be an enterprise B2B model. It’s kind of a similar model to that of the developer, but the person both developing the application and consuming it is the enterprise itself.”
To some, the thought of collaborating with hyperscale cloud providers might undercut the rationale for developing edge services in the first place – namely, to take back some of the control and monetisation of data storage and processing from them.
“If you look at operators today, they’re doing a lot of work trying to find a middle ground where they can still get their five nines but find a way to deliver applications around it,” he said.
In particular the GSMA’s initiative, the Telco Edge Cloud, is a collaboration among many major operators that interconnects their edge compute capabilities. This would overcome one of the major disadvantages that operators have typically had compared to the hyperscalers – a lack of global reach.
However, there is a pragmatic argument to be made, as Mishra pointed out: “The developers have chosen their platform – that is, cloud – and we in the telco world will have to somehow work around it.”
Indeed, there is evidence of a more collaborative approach towards hyperscalers by several operators. Mishra noted: “Verizon, for instance, some of whose edge compute is on AWS Wavelength.”
“So eventually I think there will be a middle path where the applications which are telco-centric will continue to run on the telco’s infrastructure – but not all of them, as we have already seen several operators taking a cloud-native approach to network functions. Meanwhile the enterprise functions and industries will all have to run on cloud-native anyway.
In with any outcome, the need for partnerships to develop and host edge applications adds additional levels of organisational complexity, though it may well help in building adoption.
“It’s not a straightforward ‘I’ll install a datacentre and I’ll be up and running in an edge computing ecosystem’,” Mishra commented, “I don’t think that’ll happen.”
There is a bizarre tension in edge computing between simplicity and complexity, which the telecom providers will have to navigate.
On the one hand, using edge services needs to be a simple process in order to enable usage from smaller companies and developers without expertise specifically with edge computing.
Operators can learn from the hyperscalers, according to Mishra: “The cloud companies give them a ‘developer experience’ – an experience which is feature-rich, which is agile, which is simple. You can make three clicks and have your application run on any cloud in the world today. You can’t imagine that today with a telco.”
At the same time, delivering a simple-seeming edge experience will be immensely challenging, masking layers of complexity.
“A typical network would not be homogeneous. It would be a multi-vendor ecosystem with [the Radio Access Network] coming from somebody, core from somebody else, applications from third-party developers. Meanwhile, enterprises would have their own enterprise architectures which would need to be synched up with this whole 5G,” Mishra noted.
The enterprise architecture element is itself a critical piece, which is however outside the hands of the operators.
“The enterprise has its own IT infrastructure, its own process controls, its own automation system, procurement supply chain and so on,” Mishra pointed out. “So whatever is being run outside their network, outside their premise on the operator edge compute – that has to be part of the enterprise architecture. The enterprise architecture itself will have to go through a transformation to take the edge ecosystem also into account.”
This is where Altran comes in, Mishra said: “There must be a playground somewhere to test out the 5G network or such an architecture; test it out, validate it, and see whether there are still areas that need to be fixed.”
Meanwhile the enterprise also needs to try out the edge for their applications: “Test that infrastructure, test those use cases, test the network and whether it actually matters to the enterprise to have 5G in the first place,” he said.
Mishra pointed to their 5G Labs and particularly their Edge Labs as such a playground: “We actually think that’s a big advantage for us because we have an acknowledged competence. We’re investing in the lab. We have our own software frameworks which can create and test equipment quite easily.”
Assuming that the challenges facing commercialisation of edge computing can be overcome, what then? What would this mean for the state of telecoms towards the end of the decade? Mishra has a few controversial thoughts on this.
“Connectivity isn’t going to be an exclusive domain for the operators alone,” he suggested, which would be all the better for the operators.
“If more and more industries start adopting connectivity as a possibility – whether meant for productivity or creating new revenue sources – it’s good for everybody. For the hyperscalers, for the enterprises, [infrastructure investment] starts to make sense.”
Alex Lawrence is Managing Editor at 6GWorld. His mission is to bring together stakeholders from across industries, countries and disciplines to make sure that, as technology evolves in the coming decade, it’s meeting the changing demands of society, government and business.
He has been involved as a professional nosy person in the telecoms sphere since 2004, with short detours through industrial O&M and marketing.
If you’d like to talk to Alex about your ideas or projects he’d love to hear from you. @animalawrence or firstname.lastname@example.org.