Guy Daniels, TelecomTV (00:24):
Hello, you are watching the Next-Gen Telco Infra Summit, which is part of our year-round DSP Leaders Coverage. I'm Guy Daniels and today's discussion looks at building a future-proofed telco infrastructure. No easy task. As we will find out, there are a lot of factors involved. So we are going to try and identify the key considerations and strategies for building telecom's infra that will provide a platform for sustainable growth. And I'm delighted to say that joining me on the program are Brendan O'Reilly group Chief Operating Officer for Building Networks, Beth Cohen, SDN Network product Strategy at Verizon Business Group. Fernando Castro, Cristin VP and GM Telco Infrastructure at Hewlett Packard Enterprise, and Sree Nandan Atur Technical Director Cloud with Rakuten Symphony. Hello everyone. Really good to see you all. Now we're going to be covering a range of issues pertaining to Telco infra, and there's a lot of ground to cover here. I'm certain that our audience will send in additional questions for our live show on areas that impact their roles and their requirements, but let's start off by asking you how do telcos ensure that they align their infrastructure expansion with the actual demand roadmap? How do they balance their priorities? Sree. Let me come across to you first for your comments.

Sree Nandan Atur, Rakuten Symphony (02:08):
Yeah, so first off, automation plays a really important role in managing the network. It gives you enhanced agility, faster time to market, and also it's shown that automation cuts down the operational expenses by up to 30%. Now telcos need to adapt, adopt open telco infrastructure, which gives enhanced service rollout and reduces the total cost of ownership while giving an environment for fostering innovation and vendor flexibility that telcos also have to adopt platform mindset moving away from siloed operations to collaborative networks, and it has to pay attention to the customers for continuous feedback, loop for the pain points requirements to drive the features and the roadmap to directly go to production. And since the network is built on cloud native, the infrastructure expansion would actually happen incrementally not overhauling the entire network, and it has to make sure that there are parts of the network before going into expansion that the parts of the network are not underutilized or over provisioned.

(03:14):
And predictive analytics plays a very important role here, which means that there is actually a need for infrastructure expansion to keep up with the demand and the strategic roadmap has to be such that it has to consider the customer regulatory and enterprise needs to make sure that the investments are not only happening for the 5G expansion but also for the edge investments, which are very important. And finally, the infrastructure expansion has a significant CapEx risks because the CapEx has to scale along with the revenue growth and in order to offset some of these risks, there has to be collaborative partnerships with the hyperscalers to adopt pay as you grow models. And these are some of the key initiatives.

Guy Daniels, TelecomTV (04:02):
Great, thanks Sree for going over those points and those priorities, Fernando, let me come across to you next. How do you see the needs and how should telco's balance their priorities here?

Fernando Castro Cristin, HPE (04:16):
Well, I think that Sree mentioned the most important things around the scalability, the flexibility that is needed, but also the security that is needed into those networks. So yes, we can abstract the technology to make it easier to be managed. We do have a lot of experience at Get Back Enterprise within the telco and what we are able to do is also to involve all the partners, all the ecosystem like the network equipment providers that play a key role and the operators to build the right technology in order to simplify this network evolution and basically to move the operators from managing a cost and managing a network cost to be more efficient, but also to move them into generating revenue by having the capability to deploy applications in a much faster pace than before. So that's our focus today. This transformation is not easy, it's complex and we're here to solve the equation.

Guy Daniels, TelecomTV (05:25):
Thanks very much. Fernando and Brendan, what are your views here from your perspective? How do we manage these priorities?

Brendan O'Reilly, Boldyn Networks (05:34):
I think the biggest challenge for all network operators is they're trying to balance four things. They're trying to balance coverage, capacity and capability, but ultimately cost and cost is the thing that's really top of mind at this moment in time because we're looking at the financial performance of the MNOs and how that's trending over time and cost is obviously the key for them moving forward.

(05:59):
I think the real challenge for our customers is that they're trying to continue investment while balancing costs and what neutral host providers give them is the opportunity to expand their networks, whether that be coverage or capacity, a cost base that's more attractive. What we've seen over the last couple of years and moving into this year is their desire to ensure that they do continue to deliver for their customers. They absolutely all have their customer's best interests at heart, but the challenge of 5G and the lack of a silver bullet in terms of what that use case might be, we're seeing them look more at their cost base and whether that's neutral hosts or net coasts, that's absolutely a key focus for them over the last couple of years and into this year.

Guy Daniels, TelecomTV (06:42):
Great. Thanks Brendan. That's really interesting. Beth, we've heard from our other panelists here about the considerations that Tokos need to be aware of whilst balancing these priorities. What's your view? What are your opinions here?

Beth Cohen, Verizon (06:58):
I got on some of what Fernando said and also what Brendan said. So Fernando mentioned the cost reduction and cost reduction is certainly part of it, but I think that the other major thing is regulatory issues and we still don't a hundred percent, and I think Brandon mentioned this, we still don't a hundred percent understand how the network is going to change as customer's requirements change and we've seen that already during the covid disruption if you will. So many customers changed where they were working, how they worked, the network needed to respond to that and it needed to be essentially reconfigured to support the new traffic patterns. And we're seeing that for sure as traffic patterns are changing over time that we need to respond to that. Another factor I think, which I don't think anybody else has mentioned is risk as certainly many of the telcos and Verizon's not alone in this, has put in a lot of capital into buying spectrum for example, and that comes with significance and so we need to make sure that our bet, if you will, on the spectrum does pay off. And right now I think we're in a shifting time and we're not a hundred percent sure that everything is going to get the return on investment that we expected. I do know that there's been a number of telcos have sold off spectrum and moved it around. So there are definitely things happening as we go forward and as we move to create the new infrastructure to support the new use cases.

Guy Daniels, TelecomTV (09:14):
Great. Thanks very much Beth, and thanks everybody for those opening comments and as Beth just said that it's very difficult to predict future customer needs. Which leads me onto our next question here. Is there actual evidence that enterprise customers really want access to a high capacity, high availability data center-based infrastructure with edge architecture and this strong interconnection capability, all the bells and whistles is the high investment associated with this strategy actually justified? And Brendan, I'd love to put that to you first.

Brendan O'Reilly, Boldyn Networks (09:52):
I think what we've seen from enterprise customers is that investment is there if they can associate it with an outcome for their business. So the customers we've been working with from a coverage as a service or private network viewpoint absolutely are willing to invest as long as they can take customers and shareholders and show that there is an associated value that they're driving towards. Now whether we get to the ultra reliable, low latency edge computing piece, I think there are steps to be made between where we are today and that, but if we as providers to our customers can help build the case that shows the value so far we've seen that customers are absolutely willing to invest.

Guy Daniels, TelecomTV (10:38):
Great. Very promising to hear that. Brendan and Beth, I'll come back to you because you picked up on the investment and the risk factors on your previous answer. What do you think is this strategy actually justified?

Beth Cohen, Verizon (10:54):
I thought some people are rethinking it. There are absolutely use cases where you need that high bandwidth, low latency broadcast media, not so much streaming but live broadcast. You certainly need that gaming, augmented, that kind of thing. But we have a whole lot of enterprises that have uses that they actually don't care for 5G at all and they're not looking for low latency, they're not looking for the high bandwidth iot for example, the majority of iot are sensors out in the field. They don't need low latency, they're not interested in low latency. Most of the sensors like 99 plus percent of sensors today are still running on 4G and our customers have shown zero interest in changing that. So the revolution on that side of the house is going to be much slower. So it's very much dependent upon specific use cases.

Guy Daniels, TelecomTV (12:10):
Thank you very much, Beth. I'm sure we will follow up on this on our live q and a show, so if any viewers have got questions or viewpoints, then please do send them in. Brendan, let's come back to you for some additional comments here.

Brendan O'Reilly, Boldyn Networks (12:22):
Yeah, I think Beth makes a great point, which I think aligns to what I was saying, which is people are willing to invest if you can give them the outcome that they're looking for and whether that's wifi 4G, 5G, it's imperative that we help build the case for them that drives to that outcome. And I think we're not selling technology. I say that's probably been the case for the last couple of years from the telco point of view, what we're selling as outcomes and if we focus on outcomes, there are enterprises out there that are happy to invest to drive shareholder value.

Guy Daniels, TelecomTV (12:56):
Absolutely. Thank you Brendan. Well, we are going to move on to some technology areas now because I'd like to next look at the role of abstraction here and ask you how important is the need to abstract the technology and network infrastructure layer from the services and the external customer facing view? And if this is something that's desirable, what needs to be put in place in order for us to do this? And Sree, I'm going to come across to you first please.

Sree Nandan Atur, Rakuten Symphony (13:29):
Yeah, so abstraction plays a very important role in managing the network. So let's look at the network from a 10,000 foot view. So what you would see is mainly four different components. One is the transport, second is the infrastructure, third is your cluster, and four is your applications that run on the cluster. Now there's so much variance in these four components because the transport consists of your routers, switches, firewalls and controllers and the different configurations. It can come from different vendors. Infrastructure is commercial of the shelf servers coming from different vendors coupled with hardware resources and bespoke specialized hardware accelerators, their tools and their firmware. And if you look at the software stack, there's operating system vendor, Kubernetes platform vendor, cloud native platform vendor, and the applications themselves, because it's telcos, the hardware and software is completely decentralized into different edge networks in far edge, edge and core.

(14:31):
And they have to be able to interpretate with each other now

(14:35):
since there is so much chaos, abstraction absolutely makes sense at the top and it gives you consistency with consistency, it gives you operational efficiency which gives you cost savings. There's a possibility for automation which gives you enhanced agility, time to room for innovation and all the good things.

(14:52):
Now, the way you would do this is actually build a centralized orchestration platform that looks at the network from an abstraction point of view where everything is software defined and the entire network is programmable, which means that it used standardized APIs to interact with the four different components and many different layers underneath to actually be able to program each one of them to do all the lifecycle management operations like bootstrapping, provisioning, healing, scaling, configurations, upgrades, and manage the entire network not only for day zero but also for data operations and also enable monitoring and analytics so that you can keep the network coming along, keep the applications healthy, the network healthy, and maximize customer satisfaction.

Guy Daniels, TelecomTV (15:40):
Great. Thanks very much Sree. And we've got a couple more comments on this point. So Fernando, I'll come across to you next.

Fernando Castro Cristin, HPE (15:47):
Yeah, sure. So Sree has before mentioned, exactly the equation of complexity of the different components of a network.

(15:59):
If we don't have something that centralized and enables the management of the overall infrastructure as one, you will have to go through consolidation and coordination of multiple different management platforms addressing specific silos, and that's not going to be something that allow the operator to have this flexibility, this scalability that we were talking about.

(16:27):
If you want to provision new services to deliver, like Brandon said earlier, new capabilities to a new outcome for an enterprise, you need to be agile. Agility means capability to do it fast, to do it robustly, and that's why you need this abstraction. So not big differences to what SRI was said, just confirming that's exactly our view and that's why we do have those solutions and we build those solutions to be able to deliver outcomes for the customers.

Guy Daniels, TelecomTV (16:59):
Great. Thanks very much for those viewpoints. Fernando, I'll come to Brendan in a moment, but Beth, what's your view on the role of abstraction?

Beth Cohen, Verizon (17:10):
I think abstraction is extremely important. However, one thing that hasn't been mentioned is that

(17:15):
we have to integrate with older systems, BSS and OSS systems in particular. So there's a lot of infrastructure that surrounds the network infrastructure, if you will, that has to be integrated with and that's not an easy task.

(17:34):
And there are places where decisions that might've been made maybe five or six years ago have ramifications that feed down the line. I'll use an example of as companies build out their 5G networks, they're frequently have running in parallel 4G networks as well. So a decision to change the billing system so that the 5G network is associated with one billing system and the 4G network is associated with the second billing system means that the companies are running parallel. Billing systems can have ramifications in terms of existing customers that need to move to the 5G network and all of a sudden are being told, well, we need to change your contract so it can have customer experience ramifications that frequently have or unintended consequences. So yes, abstraction is extremely important, but it can actually cause problems, particularly as you're integrating with older systems.

Guy Daniels, TelecomTV (18:53):
Thanks Beth, that's very interesting to hear. Brendan, let me come across to you and get your viewpoints on how we can use abstraction.

Brendan O'Reilly, Boldyn Networks (19:03):
I feel like Beth and I might be kindred spirits when it comes to some of these questions.

(19:08):
I think the complexity is absolutely the key part of this. I don't think anybody argues that abstraction is vital for networks as we move towards the scalability that we all hope for the telco operators and those looking at active networks, the challenge is when

(19:23):
the complexity that we bring in to these systems via the process of enabling the abstraction, it shouldn't be underestimated. And I think instead of in the past maybe we're selling a concept or a technology, really I think now what we have to do is build a timeline what needs to be done by when so that when the scalability is there, when the investment is needed from the telcos, they can absolutely meet the needs of their customers

(19:57):
is vital. And I think really for me that timeline is vital for us to be looking at over the next 12 to 18, 24 months because I don't think anybody denies the value that it will bring, but it's just about the cost associated with it along the journey.

Guy Daniels, TelecomTV (20:15):
Great. Thanks very much Brendan. We should pick this up in a year's time and see how we're all getting on here. A tight timeline for us all in telecoms to adhere to. I'd like to move on though from abstraction to looking at the underlying technology here and ask you whether or not you think that new telco in infra deployments will continue to be led by general purpose hardware and software or specialized bespoke hardware and software. Are we going to be seeing any changes here and if so, why might that be? Fernando, let's get your thoughts.

Fernando Castro Cristin, HPE (20:53):
Well, specialized hardware will still be there and they will be needed, that's for sure. And especially when we talk about low latency, very sensitive radio or with optical transport, those specialized hardware is the most efficient today to address those needs. On the other side, we heard from Beth with Brandon, with SRI that the flexibility that you need to have in order to operate and to deploy new services moves some of the workloads, some of the applications to general purpose. Some of the advantages are of course the orchestration that you can have through different applications, but also the pace of technology evolution. When you come to more general purpose compute, which have those applications that can deliver those services, you can have a higher rhythm of technology that comes with benefits like lower power consumption, new technologies that are introduced in a much faster pace. So those technologies that are becoming with a higher rhythm than the traditional specialized hardware, well it's a benefit that you cannot afford to refuse if you want to have this evolution of software defined networks that can deliver fast applications and new applications and to address like branded said earlier, new outcomes for new customers.

(22:31):
So it's a battle of having the right infrastructure that is flexible enough with the rhythm of technology innovation that allows you to deliver faster application. So yes, this is a hybrid world technology traditional possibility with their own software will still remain super important and will deliver a lot of the services that the operators need to deliver. But the new technologies with computer general purpose, computer technology will enable new services in a different pace and that's what I believe is going to shift step by step the infrastructure to compute based.

Guy Daniels, TelecomTV (23:21):
Great. Thanks very much Fernando. As you say, we've got to get the balance right and you mentioned orchestration there, which leads me onto a final question for all of you. I like to ask what is the benefit of having a centralized service orchestration platform for end-to-end resource management and is this a good idea to replace individual orchestrators in the network? Beth, can we come to you for your thoughts on orchestration?

Beth Cohen, Verizon (23:53):
Orchestration is like standards. There's so many to choose from, so I think yes, that's the north star to have a centralized orchestrator. However, I think in all practicality it's just not even possible to get there.

(24:11):
Certainly getting the number of orchestrators down is a good idea, but there's even regulatory issues that enter into it. Verizon has a wire line and a wireless networks, they're actually two separate networks and they're two separate business units. It's great in theory, but I think in practicality it's just not even achievable.

Guy Daniels, TelecomTV (24:39):
Great, thanks Beth. I was really optimistic that we might have a really simple solution to this one, but it looks like reality could be a lot more complicated. Sree, let's come across to you next. Do you shared a view that there are a lot of orchestrators out there and it might be mission impossible to get it down to a single orchestrator?

Sree Nandan Atur, Rakuten Symphony (24:59):
Yeah, so let me give you a different angle to this. Let's say there is a single Kubernetes cluster that you would actually want to manage. Now the Kubernetes cluster consists of control plane and the data plane and the way control plane is designed is a tight loop and it manages a state machine and it's actually based on level based triggers. That's the design and it's able to manage all of the events. Now when the applications run on the cluster, it's actually a pod and the pod goes through let's say 10 different events and a hundred nodes, a hundred pods per node is about a thousand. And let's say the average size of the cluster is 50, so you have 50,000 events, it's a cakewalk, a single Kubernetes cluster can manage. Now you'll have a single tool that can manage talk to the API server and it'll be able to manage the cluster.

(25:46):
But in reality, the telco networks are very complex and they're actually designed, the hardware and software are disaggregated and they're designed based on the geographical region because it's just physics, it has to span the entire geographical region and you need to have many, many different types of these clusters that are actually deployed that need to interact with one another. And they're actually divided into, because of the disaggregation divided into different edge networks, you have the four HHN core and these networks are connected with one another using the front hall, mid hall and the backhaul. Now that means that there are thousands of these clusters that are distributed in the entire geographical region. Now, would you have a single automation tool per Kubernetes cluster? That would be way too many. Having one is to, one is just not scalable, but how would you do? One is to end.

(26:41):
That also does not fit into the design because the applications are dynamic, they span different Kubernetes clusters. The data needs to flow in between. You need to gather the data, the monitoring analytics has to happen. So while it is ideal to have a central orchestration platform to manage your entire network, as Beth was saying, practicality and business needs might be that you would have quite a few, not like it is ideal to have one, but you would have quite a few that would actually manage your entire network. And that would mean that it would use open programmable software-defined networks to manage your entire network and give you orchestration capabilities and manage its entire lifecycle.

Guy Daniels, TelecomTV (27:22):
Interesting. Thanks very much Sree for those comments and Fernando, we'll come across to you and what are your thoughts about the future of orchestration? Well,

Fernando Castro Cristin, HPE (27:32):
I think that based on Beth and Sree's answers, you see the experience behind there is an ideal world and there's a real world. So in the real world, I think that centralized orchestration, it is possible if you define centralized as multiple centralized locations or areas, you can do some service orchestration for some domains and for some other ones it's not going to be easy enough to be relevant and for all the network as one, yeah, the network is too complex. It addresses two different domains, two different areas geographically or even from a business standpoint. Like Beth was saying that you cannot say that I have one network, you have multiple networks. It's a little bit like the universe, multiple galaxies and multiple solar systems.

(28:27):
You have to be able to go down to what is manageable. That will be your centralized service orchestration for that area. And so that will create multiple ones. The important I believe is to have a strategy around it and the strategy of those multiple ways of managing have to be consistent. So it's manageable, but one single. Yes, it's utopia.

Guy Daniels, TelecomTV (28:53):
Yes, utopia, indeed. Fernando, thank you very much for those comments and views. And Brendan, let's come across to you for your thoughts on the role of orchestration.

Brendan O'Reilly, Boldyn Networks (29:08):
I think the last two questions are sort of linked and I think it's a question we've been asking as an industry ever since, well, certainly since I joined, do we want to go to general purpose computing? Do we want well on service Orchestrator? The answer is yes. We would love this to be cheaper, more manageable, better for our customers. But I agree with the comments from everybody on the panel.

(29:28):
It is a utopia. It is hard to do. The great thing about our industry is we keep asking what if, so what there was one, what if we could go all general purpose computing, what would happen if we drive network abstraction to the nth degree? And it's only by asking those questions and by continuing to push, do we get to a point where we absolutely can see what can be done? I think the great thing about this industry is it just doesn't give up. And I think as we continue to ask these questions, we'll continue to make progress.

Guy Daniels, TelecomTV (30:03):
Thanks, you said Brendan. Thanks very much for that. You are absolutely right. This industry continues to ask questions and hopefully we'll play a little role here in help answering them as well, but we must leave it there for now. I am convinced we'll continue this debate during our live q and A show later. But for now, thank you all for taking part in our discussion. If you are watching this on day one of our Next-Gen Telco infra summit, then please do send us your questions and we'll answer them in our live q and a show, which starts at 4:00 PM UK time. I'm sure there are many questions we still need to address, so let us know what you want to hear discussed. The full schedule of programs and speakers can be found on the telecom TV website and that's where you'll also find the q and a form and our poll question For now though, thanks for watching and goodbye.