Tony Poulos, TelecomTV (00:17):
Tony Poulos here at the DSP Leaders World Forum and a new version of Extra Shot. Boy, we're getting through a lot of extra shots today and luckily I've had a lot of coffee to carry you through all of these. We've just been through a really interesting session, I think is the best word for it. Talking about the future of RAN and building the future with an open virtual and AI enabled ran. We've been talking about this for quite some time. One of the things that did come up today, and I'm probably going to start with you, I've got Robert Curran, of course, from Appledore Research with me and Martin Halstead who's the senior distinguished. Oh, the Senior Distinguished Technologies at Aruba Telco Solutions, Hewlett Packard Enterprise. That's a really long title.

Martin Halstead, HPE (01:00):
It is, unfortunately, yes.

Tony Poulos, TelecomTV (01:02):
Wow. Far too long for somebody like me. Now listen, let's start with you Martin. Rob Joyce from oh two was talking about how he wouldn't be buying virtual ran until about five years time or totally buying Virtu in five years time. What did he mean by that?

Martin Halstead, HPE (01:18):
So I think what he meant was that it's inevitable that open ran, so V ran and then open ran are going to be the predominant technologies in Telco moving forward. The shippers sailed that is going to happen. The challenge that they have, which is where the five day figure came up, is, well actually there is an awful lot of legacy previous generations out there deployed in base stations, powered issues, existing contracts, et cetera that they have with the existing incumbents that are there. So therefore when can you change? So it's not like it isn't going to happen, it's just practicalities of when they can go to those open technologies.

Tony Poulos, TelecomTV (02:06):
Right. Would you get the same feeling from that, Robert?

Robert Curran, Appledore Research (02:09):
Yeah, I think it's clear the direction of travel is clear, but these practical limitations, one of the issues for RAN deployment in general, little on open ran is the cyclical nature. It's a big CapEx investment cycle and you're not going to replace that every two years, so you've got to kind of get the timing right. I think now people, one of the concerns about open ran from the open ran community had been the risk that it got attached to something like six G, in other words pushed a long way into the future. In fact, that's not really quite the case, but the concern was there. I think there's two different things we're talking about. We're talking about if you like, the politics of open interfaces and supplier diversity, particularly on radio units in Fronthaul, and the second thing is virtualization, the cloudification of ram, they're two very different things. Sometimes we can conflate the two, but they both have different issues around them, different constraints and different business objectives at the end of the day.

Tony Poulos, TelecomTV (03:07):
Yeah, I think my sympathies went out when I heard that some network operators still running two G, 3G, 4G and 5G, and now talking about virtual RAN and AI enabled in the future, boy, how do you manage such a vast amount of legacy systems on big networks, particularly something like Verizon across the whole of the us? How are they going to cope with that?

Martin Halstead, HPE (03:29):
Yeah, so I think the challenge is, and especially with previous generations, previous generations of radio networks were monolithic, right? They came from a single vendor. The entire stack is proprietary to that vendor, and so those systems are being gradually phased out. I think in order for the entire industry to migrate to cloud native based architectures, that's already happened within the core. The reason why it's happened within the core is because it's easier to do it there, right? Obviously it's more challenging in the ran. That's why it's taken longer in the core. You already have a move originally with NFV network function virtualization, a move from monolithic applications being moved onto virtual machines on industry standard service and then a move from those to cloud native. So open architectures, microservices, et cetera. The operators love that, and by necessity are going to have to move to those same architectures across the entire telco landscape. In order to do that, you have to get rid of previous generations that aren't capable of going to those microservices cloud native architectures. But as the generations move forward, the industry is going to get there.

Tony Poulos, TelecomTV (05:04):
That's presumably a very expensive exercise as well. How are they going to accelerate the move to open ran and virtual rans?

Robert Curran, Appledore Research (05:13):
I think it's a bit of push and pull, to be honest, Tony, the pull is towards the greater agility and so on that they need for their customers. We've heard some people talking about moving at the speed of the business and tega relative ability or inability to do that. So that's the pull the push side. I think it's just some of the network economics, the cost of power, the cost of maintaining sites, having to go and do things physically on sites. If you've got 20, 30, 40,000 sites, that's an enormous amount of effort. So I think it's a combination. That's what we'll combine to accelerate the transition. You still got to work within the CapEx cycle, so that's a pretty hard constraint. I think nobody wants to depreciate something any faster than is already on the books, but there are these other factors, hey, push and pull will give them a rationale to move to more cloudified architectures.

Tony Poulos, TelecomTV (06:05):
Well, I'm presuming that the cost savings are one of the big drivers for the business case of moving towards open and virtual assessment, but what are the real cost savings? Do we know yet? Moving to open virtual

Robert Curran, Appledore Research (06:16):
Ran, certainly one of the things that was mentioned on the panel today and certainly we found in our research is it does vary. Telcos are pretty cagey when it comes to it because they're all kind of different. They have different circumstances open ran and across the country like Indonesia, completely different compared to Luxembourg or Belgium. So the topography is a factor in this. It's quite difficult to maintain it, but if you look at what people like tell us have been very bullish about tco, lower TCO, that wasn't why they did their transformation, but having done it, they reported pretty positive results from that. So we do hear a mix of different things, different experiences around the world, but I think most operators have been putting into context it's not, don't do open RAM for the TCO benefit, do it for the agility benefit, do it for the power benefit. There are other factors that make the business case it does. TCO on its own may not stand

Tony Poulos, TelecomTV (07:09):
Up. Are we seeing greenfield operators going pure virtual and open ran for their system?

Martin Halstead, HPE (07:16):
Yeah, absolutely. We are, and I think it's inevitable as not just for the greenfield operators, but for the operators that are moving to next generations of radio. They are going to be open, they are going to go to open microservices based architectures, and some of that's even going to be dictated I think by new standards. So for example, six G is going to be entirely microservices based. You will not have a proprietary stack on top of which you are going to build from a single vendor, top to bottom, a fully integrated offering as you did get in previous generations. The market has moved on and it's inevitable that openness is going to move on from there too.

Tony Poulos, TelecomTV (08:09):
And presumably all software driven as well, which brings us to the next generation, of course, would be AI enabled brands as well. I mean, I'm a bit frightened to hear that already because as Yago said from Verizon, we've been working with AI for years in developing the RAN and running the ran, but having it completely AI enabled, I think it's a bit in the future still.

Robert Curran, Appledore Research (08:33):
We look forward to it. I think the reality is there just is immense complexity and I think people are becoming accustomed to the idea of using AI to manage that complexity in real time, the amount of configuration parameters and so on. And we've been doing this in different ways for 10, 15 years. So it's not so alien in the ram. It's not such a crazy concept, and so I think there's a bit more acceptance of that. I think where I think it's getting a little murkier is some of the discussion around reusing the RAN as kind of mini AI data centers. That's a little bit more speculative, if I thought of a better word right now.

Martin Halstead, HPE (09:10):
Yeah, I think that's a great point in that the management of the RAN and actually new architectures, especially open ran where you could have third party vendors introduce ai, it's much easier to do that, to have innovative new vendors doing AI for managing, using ai, for managing the network itself, actually having GPUs, et cetera, built into a cell tower.

Tony Poulos, TelecomTV (09:39):
Don't go there. We're going to be talking for another half

Martin Halstead, HPE (09:40):
Hour. Yeah, jury's kind of out as to the benefits of,

Tony Poulos, TelecomTV (09:45):
We actually have run out of time on this particular one, but I thought we were just starting to warm up. Well, Martin and Rob, thanks very much for being with me today. And for those of you at home, if you do miss any of our sessions of the main program and even the extra shots, you can view them online I think from tomorrow onwards. So keep an eye out for those. We'll see you later today for another extra shot. Thank you.