Note, this article, a direct transcript of a conversation with Kathiravan Kandasamy, vice president of product management at US-based carrier services company Syniverse, continues from a previous entry, which can be found here.
…
You mentioned also that, in your experience with customers, that the private networks market, while presented initially as an all-edge setup for machine comms, has morphed more strongly into one for office campuses and the like, where the use cases are driven by basic consumer-style connectivity for smartphones and tablets. Just say a little more on that, and how and when you see the private 5G market developing into something else – for more profound Industry 4.0 applications.
“I mean, the thing we are finding with customers today is that most use cases, whatever the circumstances, can be met with a private LTE solution. And we ask them: ‘Do you really want to invest so much in 5G? What is the driver, really, to go to 5G?’ And we have not encountered a use case where the workload is so intensive or the latency so acute that you need 5G – or where it is complex enough that you need some sort of slicing. We just aren’t there yet. Really, in the past 12 months, and maybe even in the next 12 months, the task has been to educate enterprises about the art-of-the-possible with private networks – whether that is about high bandwidth, low latency, high security, high density, precise slicing, secure roaming; all of those things are still being played-out. Because today, the question is whether you really need a 5G core – when a run-of-the-mill 4G core can do what you need.”
But is there not a danger not-to-say that you are future-proofed against this art-of-the-possible with a 5G core?
“No, I get it. And don’t get me wrong, I want to sell a 5G core. But there is a question, still, at this stage, about whether it is worth it.”
So how does this market develop? Would you say this ideal – about enterprise-sited, enterprise-owned, enterprise-managed high-fidelity private 5G networks is a pipedream?
“I would say that, at the moment, most enterprises want to get their hands on 5G, and build their own RAN, and manage it all themselves; but that this will fizzle out over time. It will end up becoming more commonly a hosted and managed service – meaning they will use someone’s core, and someone else will run the RAN and core. Because it is not their core competency. And because, in the end, the cost will be less with hosted solutions, which are multi-tenant.
“The vendor providing the solution will make it carrier-grade, fully-robust, with 100-percent redundancy and five-nines reliability, and so on; just at a fraction of the cost, because it is a hosted solution. It could be in a public, private, or hybrid network; it doesn’t matter. Like I said, we have so much security control today that if you don’t want anything to go onto the internet, then we can do it. Not a problem.”
So is the correct way to look at this that the market at the moment, when it talks about this stuff, is obsessed with a future where Volkswagen is animating its production lines using Release 18-level URLLC 5G networks? But that the vast majority of the wider industrial market does not care about all of that? That this VW ideal is a drop in the ocean?
“I believe so, yes. It is not just industrial markets. We are banking on 50 or 60 billion IoT devices; every device is going to be connected, right? So when every device is connected, there will be a serious need to apply and manage policy and security on top of all of this. You can’t run everything on islands. Because it just defeats the purpose of ubiquitous connectivity. It breaks the model, right? So it will be that hybrid model, in the end – with this level of security for these types of use cases, and with pervasiveness for devices to be here and there, and anywhere. We are even thinking about interoperability between private networks. Because it’s going to happen.”
So it is almost like the future of private networks is not private networks. The future of private networks is somewhere in the middle – a network of radio networks with shared core infrastructure?
“The future is going to be hybrid networks, absolutely. And we are also following the WBA with Wi-Fi 7, and where that goes; and again, from a performance perspective, they are decreasing the gap with 5G, right?”
So what about the argument that, you know, HPE has bought Athonet and Siemens has its own OT-grade 5G system which will be sold as a private network, and factories will buy and deploy cellular in the same way they deploy Wi-Fi? That it will not use centralised cloud infrastructure, often; it will be on-prem, and it will become such an easy box-sale that you can almost run a dealer channel to sell it to enterprises?
“Yes, I know, that is the dream. But I also think a lot of people oversimplify the complexity of cellular. Because if you oversimplify the management of the radio access and spectrum management, then what do you lose ultimately? Some security? Some manageability? The reason enterprises like this is because they want the flexibility that carriers cannot afford them today. Because when enterprises ask a carrier for this number of IMSIs with this type of data plan and this type of performance, the carriers say, ‘I’m not going to change things just for you’. They have five airtime plans, seven data bundles – you can take it or leave it. Which is why there are companies like Syniverse in this space – which want to simplify and to extend all of their features and functions to enterprises. But you want to shrink that into a box? It would be like trying to shrink cellular into a set of APIs.”
So I keep repeating this story; but Volkswagen was on stage at Hannover Messe (in April), complaining about 5G, and all the problems it has with it. And it said it has to work like Wi-Fi, and that, in the end, the SIM has to go; that the authorisation and authentication processes for industrial 5G devices have to be the same as with Wi-Fi. That it cannot be a SIM-based system anymore, which sounds like a big fundamental change.
“It is, and it goes to the question: do we still need a 10 digit phone number? I think we are still way off that, today. We are not there yet.”
Are you saying we don’t need a 10-digit SIM-based number, and that it is a long way off? Or are you saying we do need that – we do need that SIM identity and carrier grade access and security?
“No, I am not beholden to the form of authentication that we have today. If the question is why you need a physical SIM, then that argument is already gone. We already have eSIMs and iSIMs today. It’s just a form of authentication, which you can roll into your AAA (authentication, authorization, and accounting), or whatever radio set that you have. But there are also some legal aspects; like for Wi-Fi authentication, there is no need for anyone to really know where the device is or who owns and operates the device. But through legal contacts in many countries, the operator needs to know who has that IMSI – whether it is tied to a person or a company, for instance. Because of the devious things you can do with a cellular device, right? So there are also some legal constructs behind this.”
But that also suits you, right? I mean, your business is based on that model, so you see value in that?
“Correct. A lot of the feature-functions we have built are based on the IMSI. Because we know which IMSI the device is using, which is how we actually apply that control. And IMSI is unique.”
So is the argument that you cannot simplify cellular to Wi-Fi levels? That cellular is by its nature a complex beast, and it will not be box-shifted like a Wi-Fi system? And that it will be centrally controlled in the end – that the original centralised model for cellular will remain to a greater extent?
“Yes. I believe so; in the near term, absolutely. It is not going to change even with 5G.”
