Guy Daniels, TelecomTV (00:12):
Hello, you are watching the Cloud Native Telco Summit part of our year-Round DSP Leaders Coverage. I'm Guy Daniels, and today we're going to take a closer look at containerized applications and cloud native network functions, CNFs within the telecoms industry. And joining me now is Francisco-Javier Ramón Salguero, who is multi-cloud tools manager at Telefonica, and also the chair of the Etsi Open Source MANO initiative. It's good to see you again, Francisco. Javier, it's been a while since we last spoke. A recent report came out early this year from IDC suggested that we are now finally seeing a tipping point in the migration of virtual network functions VFS to CNFs. Do you agree? Is this something you see as well?

Francisco-Javier Ramón Salguero, Telefónica (01:06):
Well, I would say that that matches the perceptions. I mean, it is true that there is a growing confidence in that code that CNFs can be suitable for filling many of the use cases. I would say vast majority of the telco use cases. And also there is a growing maturity in the offers. So that is helping on that. You see that there are also advantages. I mean if you are adopting A CNF in comparison with the BNF, it means more predictability, less trouble, more clarity on the application life cycle that is good for the long run to understand how to manage and how to scale that piece of software. And also the things that since we are having more offers, we are gaining more experience. I mean we are also gaining more confidence is that the next deployments if they can be, if the CNF is a suitable option, that will be the preferred option in principle. So I tend to agree with that perception, perhaps not in the figures because the estimations are too complicated. But in general I have the feeling that this is the top one option for new deployments.

Guy Daniels, TelecomTV (02:30):
Great. Well thanks very much for that. And we know that these organizations do like to produce their figures and forecast. I've got another reference here, another report, this one from Gartner that says in about three years time or so, or 2028, more than 95% of global organizations will be running containerized applications in production environments up from what they estimate to be about 50% last year. So what's behind this increase? You mentioned on your previous answer that you're seeing maturity in CNS, but what's behind all of this?

Francisco-Javier Ramón Salguero, Telefónica (03:08):
Well, I think that this, just to put things in context, I mean telco is just another vertical for virtualization and we have seen this trend already happening in other industries. So in telco we are following with some delay because the peculiarities of the type of workloads that we are managing and the ecosystem that is quite special. We are seeing that trend emerging a bit later. But it is not surprising if you analyze in IT industry in general in different verticals, the adoption, I mean you are seeing that Kubernetes is everywhere. I mean there is a large ecosystem around Kubernetes, around the CNCF umbrella. That means that there are components that are designed specifically to be added on top of your existing deployments to enrich what is there. So for instance, if I need to encrypt the traffic and my application is not supporting to any traffic encryption, I could have a service mesh to encrypt it afterwards. And that works. We are just seeing the tip of the iceberg in telco, but this is behind the scenes. There are a lot of potential that we can unlock. I'm pretty sure in the coming years we'll be seeing in all the same level of flourishing that has happened in other IT sectors, so to say.

Guy Daniels, TelecomTV (04:43):
So you already covered a number of key telco use cases there that are now fully supported by CNS, but I'd like to perhaps ask you where further work is perhaps needed here.

Francisco-Javier Ramón Salguero, Telefónica (04:54):
We see a lot of disparity, I mean that are designed for being interoperable with almost any type of Kubernetes infrastructure provided that it has the right type of resources and the right amount of them of course. But in some other cases there are CNF that have been designed with a very, very specific platform in mind, which is the exact opposite of being cloud native, right? We are still seeing that orientation for verticals and that other extreme, the other end where there are applications that are designed to be deployed anywhere. And I think that we need to see a convergence there eventually for applications that are designed for being well-behaved applications in a Kubernetes ecosystem. So they can coexist as I mentioned, with other tooling that you have there with your security scans, with your policies, et cetera, et cetera, without being intrusive with them still. I mean there are rough edges there in industry that hopefully we'll be solving in the coming years. It's a process and we understand that we need to go through that to get to that level of sophistication.

Guy Daniels, TelecomTV (06:13):
And as you say, it's a process and we will get there as an industry, this will come in the years ahead, but how does a telco run and manage all of this? How does a telco actually go about it? And perhaps what are some of these specific benefits of using Kubernetes for cloud native network functions? And you mentioned earlier the declarative nature for example.

Francisco-Javier Ramón Salguero, Telefónica (06:34):
I mean industry is going in a direction where you find more and more horizontal deployments as well, which have a lot of benefits because in the end, for a new piece of software, you don't need to do the low level design from scratch. You can start working with the management of software. So there is also some approaches where are developing, where you see the Moreal platforms at some point, those Horizonal platforms where you have sufficient confident, we are seeing that there are some sort of GI tops in place where with well-behaved applications with applications that do not behave that well. You need to treat them separately. And there are, in a operator rule, the 80% of the effort I'll say. But in that 20% of the effort, that probably is a larger size of the larger set of the deployments, you can start introducing GI tops practices, which simply means that having recording of all those intents of deployment, those declarations of intent that you are doing, having them recording in a control system, patient control system so you have sort of a time machine of what you've been deploying in case that you need to roll back or something.

(07:53):
That is something that we are seeing and actually we are working a lot in that direction. In the end, what we are trying to achieve, and this is one of the big promises of Kubernetes and this ecosystem about Kubernetes, is that it's giving you better robustness and more flexibility combined. So typically you have a trade off. You have a trade off. If I want to be more flexible, I need to be less robot robust because I'm moving faster. But now this technology gives you, is a leap forward in that sense and that tradeoff, the cion is somewhere else much farther and you can go much farther in that robustness and that flexibility without compromises if you do it well. Right. And there is the direction we are where we are going now.

Guy Daniels, TelecomTV (08:45):
Great. Thanks for explaining that to us. And one final question for you, and this is based on a couple of questions we've already received in from viewers for the summit and they mentioned that containers are not monolithic, the ecosystem comprises many components. Can you help explain what some of these components are and why they are important?

Francisco-Javier Ramón Salguero, Telefónica (09:07):
Yeah, I think that that is one of the keys behind the Kubernetes philosophy and containerization is that, I mean in the all wall of virtual visualization, when you created a virtual machine in the end it was a rough equivalent in virtual wall of a physical machine. And how you operated that one that meant that you will put in there not only maintaining the operating system, but also maintaining all the set of components and the different versioning that were needed for the application to run inside. And you are putting all that together in a machine like we've been doing in it for four decades. That is what changes in Kubernetes because the problem with that approach, which is not a problem because it's been working, but the challenge that you have is that the case that works were wrong in there. Were tightly coupled all the components and they need to coexist in the same operating system, the VEing, and they could be affected substantially with any patching or any upgrade that I did to one of the components.

(10:14):
And that requires to do it successfully significant knowledge about how was all that ecosystem working together. And you were quite conservative, although in the upgrade of the different companie in case that you broke something. That is what the CoreNet and the containers in general allow you to decouple. I mean, we thought of our server, we have the good old lamp, so you have the liners, you have the Apache, you have MySQL and PHP and all there were combined in specific versions in one single machine that is not needed anymore. In a container world where you can decouple, and this is the opposite of a monolith like the one that I've described. You can put the database in one place and maintaining and upgrade it at its own pace. You can put the from services somewhere else, you could put some PHP in other containers and you can maintain them in isolation. And still you're declaring how they relate each other. So you don't lose the track of what is going on and how they are relating. But each of the component in principle can be updated separately with a stronger confidence that will not affect the rest of the components. Of course, there are limitations on this approach, but it's much simpler to enter in the ecosystem and maintain continuously all the components if they are not that tightly coupled. And that is one of the big advantages of this approach.

Guy Daniels, TelecomTV (11:57):
Fantastic, great explanation there. Thank you very much. But we must leave it there for now, Francisco Javier, good talking with you again. And thanks so much for sharing your views with us today. Now don't forget to watch all of our videos from this year's cloud native Telco summit. New panels and interviews are available to watch on demand during each day of the event. Goodbye for now.