Industry EventsMarch 20, 2023

Telecom Transformation – Disaggregation, Software, Cloud

AT&T, Orange, and Telefonica on Disaggregated Cloud-native Networking

DriveNets hosted a panel for analysts and media at MWC23 in order to explain its work and hear from industry leaders on progress for disaggregated networking. Members of the panel included Igal Elbaz, SVP and network CTO at AT&T; Ido Susan, co-founder and CEO of DriveNets; Jean Louis Le Roux, executive vice president for International Networks with Orange; Cayetano Carbajo, VP for core, transport and service platforms with Telefonica.

See the full video

Short version video

Full Transcript

So just a little bit about DriveNets for those who are new to us, our vision is to build networks like cloud. And the idea of building networks like cloud is really to empower new possibilities. When you move to software, everything is possible. So the disaggregation of hardware, the software, the moving to a cloud like model, really unleash opportunities for our customers and even for us. But it’s all about our panel. So our speakers today are Yigal Elbaz, who’s the SVP and network CTO for AT&T. So thank you for joining us Yigal. So we confuse you with the way they sit, but that’s okay. On my right here is Jean Louis Le Roux, who is the interim Executive VP for International Networks at Orange, and Cayetano Carbajo, who’s the VP of Core Transport and Service Platform at Telefonica. And then Ido Susan our CEO.

Ido Susan, CEO, DriveNets
So everybody know that traditional network, that the up level, the legacy, what we call network that building and focusing from hardware based solution. They have three end customers, the Consumer Enterprise and Broadband. And this is a traditional network. As of today, we are here and we create DriveNets and fund DriveNets in order to do the transformation meaning to move those customer or those networks to be cloud. What it’s meaning? Cloud, it’s meaning shared infrastructure, open infrastructure, running multiple workloads. And more important, once you certify the cloud to add a new workload, it’s just to add another software that leverage the resource that you have in the field. So instead that you want to add more capacity and you’re running after line cards or instead of to certify another box or routers, physical routers from additional vendors, incumbent like Cisco, Juniper, et cetera, everything running on the software. So this is what we are doing. So this is what we believe, this is the involvement of the network that we believe that will happen in the future and what we are doing with our customers. So we took the same concept that the cloud did in the compute and storage, what we familiar today with AWS and Azure, et cetera, and we just replicated to networking. So if you’re looking on the compute and storage industry, it starts from the mainframe moving to X86, then virtualization of VMWare and now everything it’s cloud as we’re familiar, but nobody aware when on the hardware where is running, etc. We did exactly the same on the networking.

Yigal Elbaz, SVP Network/CTO, AT&T
Good afternoon everyone. Yigal Elbaz with AT&T. I just want to give a high view of our journey. So we’ve been in this disaggregation virtualization and software defined networking journey for almost a decade. And we’ve been very public about this. We’ve been very not about just the journey, but we put goals that we’ve announced them in public, so we had to go and deliver them. We achieved all of them in 2020, but it’s a journey that we’ve continued. However, the number one challenge, which is a good problem to have, it’s that the traffic coming from the left side is growing in about 30% year over year. So again, this is a good problem to have because that means that our customers are enjoying our services, are dependent on our services and using more of our services. And over time we had to develop an approach and an architecture that allows us to scale our network much easier than before. So a lot of this is running on concept of cloud, as Ido said, like how do you scale easily, how do you run software on top of hardware. And we’re trying to apply those principles into every part of our network. I think the last thing that I want to say around cloud is we understand that today everyone on the left side are consuming more and more content and services from the cloud. So the architecture looks different today, whether it’s consumer or enterprise. So being able to bring together users and branches into content that can be in different locations and doing this in a seamless network experience, no matter what is the transport, is something that we’re building on top of this. So with that, again, as Ido said, the part that DriveNets we’re using is our core backbone. We are carrying 590 petabytes of every day. That’s a lot of traffic. And on our core backbone. Already over 50% of the traffic is running on an open disaggregated architecture where the network operating system was DriveNets. It and we’ll talk probably about this later, but we enjoy that relationship. We enjoy the disruption. We enjoy the ability to work together and push some boundaries in our ability to deliver not just software, but actually a full service that we can deploy. The reliability and the rigor that telco at our scale need.

Jean Louis Le Roux, Executive VP International Networks, Orange
So, yes, I’m part of Orange and managing the international network transmission, iP Voice Roaming. Today we are going to focus on our international and global IP MPLS backbone. We are running a backbone that is deployed in 100 countries and roughly 250 points of presence. And we provide connectivity from a few meg to 400 gig. We provide connectivity as a tier one internet to wholesale customers, and we provide also connectivity to enterprise customers and VPN and cloud connectivity. And the traffic, I think we have the same trend. The traffic is growing, it depends between 30% every year to 40% depending on the year. During the COVID crisis, we even add in 2020 plus 60% of traffic during the year. So this is what is at stake today and our work within it. Orange international network is to make sure we can absorb all these traffic increase while, of course minimizing our cost while ensuring the right quality of service to our customers and also very important, while minimizing the carbon footprint. At the end of the day in order to cover the cost efficiency, the scalability, the green and the on demand. We need elasticity, and I really like this word elastic. We need elastic and flexible networks clearly and in order to answer to those requirements the solution is disaggregation as you clearly demonstrated. Disaggregation. Why? Because by separating the hardware and the software you can manage your network like cloud and you can scale in an elastic and in a flexible way. This is why we decided one year ago to move towards disaggregation. So in a nutshell we start very simply, we go step by step. We have already disaggregated nodes up and running in the network and we are very happy because we did the math and the TCO improvement is about 20% when we disaggregate. Now we are studying the core router and here clearly I would say that DriveNets we have a very strong discussion, very advanced discussion with DriveNets. DriveNets is clearly a very good candidate. I think the cloud network architecture is really relevant to provide this elasticity. So we are going to start doing some testing in our labs and most probably we will have a trial before the end of the year.

Cayetano Carbajo, VP for Core, Transport and Service Platforms, Telefónica
Well, good afternoon Cayetano Carbajo leading Core Transport and Service platform in the technical corporate unit of Telefonica, what we call global CTO and it is my pleasure to be here with you to comment about this disaggregation or softwareization or cloudification, you name it, in Telefonica network. So we are pushing for disaggregation in all the network layers. Why? Several reasons. First of all, when decoupling the software and the hardware you can have different innovation cycles and software cycle could be faster and you can bring new things to the network in a faster way. You are eliminating one interval area for this innovation. This is as well important for automation. Many of the automation basic capabilities are coming with the software and we have a huge need of automating this as well and a ecosystem reason. So when reducing the entry barrier of the hardware you can have more vendors, more partners for your network. So we are definitely pushing for disaggregation. As you can see we have been pushing for the disaggregation in all the network layers. In here we are about to do a swap remotely. So we are keeping the hardware and we want to change the software provider of something like 2000 routers. What is bringing to reality one of the promises of disaggregation? You can change the vendor remotely because of changing the software. What will be an achievement? I would say if I go higher in the hierarchy, what we call HL4. We are running right now commercial processes and maybe we decided to go for the disaggregated vendor for Biras and of the top for HL1, HL2, I mean Peering and Transit is where we have been considering DriveNets. We have not done in operations yet but we have been testing DriveNets and the result has been good.

Questions and Answers
Question #1
Hi. Very impressive by the way to see where you were in the architecture roadmap. One thing I have because I remember working on some projects with AT&T and comparing it to Google’s to manage a server was like $600 a year and it was $0.06 for Google. So this direction that you’re going with cloud makes a lot of sense. But have you looked at what the outcomes were for an economic point of view to say here’s what it costs me, whether it’s cost per gigabit or whatever measures and says here’s what it’s costing me now. Now I have 52% of the traffic there.


Is it close to the Google example in the past?

Yigal Elbaz, SVP Network/CTO, AT&T: We’re getting really good TCO out of that journey. Everyone need to understand we don’t do anything just for the sake that we said that we’ve first mover or we work with disruptors or we work with the startups. It doesn’t matter. We’re building an architecture that we believe serves us best in terms of scalability, cost. And we’re trying to see who are the players that can fit in and if they can fit in and we can deliver then we’re checking ourselves all the time to make sure that some of the assumptions that we had at the beginning make sense. Actually, I think in this case are actually better.

Question #2
I’m Kelly Hill with RCR Wireless News. How does this change testing and monitoring for networks is DriveNets responsible for implementing changes? Do you folks work together? Is this all still in AT&T’s wheelhouse? How do you approach network testing and monitoring and sort of making sure that everything works the way that you need it to work?

Yigal Elbaz, SVP Network/CTO, AT&T: This is a great question. We do everything, we take the hardware, we take the software, we take the subcomponents and we are doing all of the testing, we’re running all of the regression. We’re working with each one of the vendors to get their roadmap and we’re bringing this together and we’re checking all of the interfaces to other parts of the network because the core has to work with the edge, has to work with our systems. So AT&T is responsible end to end for every piece of hardware or software that gets into our network. And it’s all done by us. Clearly that allows us to move a little bit faster because we have that level of disaggregation. Not all of the components are moving at the same pace. But still we have getting software from DriveNets and we’re responsible for getting this into the network.

Ido Susan, CEO, DriveNets: I would add for that that it really depends on the customer. So we have a US based customer like Yigal mentioned that really control the entire chain and probably getting the biggest value. And we have other customers either in Japan or India that want to the first phase that we will be the one throat to choke meaning we’re working with local integrators like, I don’t know, CTC in Japan. So they’re getting the hardware from the ODM, they’re getting the software from us. We’re doing the certification. We verified that the optics and everything is working and they deliver this to the customer together with us, we’re doing the certification with the customer. So it really depends. We have the two model tier one service provider, typically taking control end to end from the service chain.

Cayetano Carbajo, VP for Core, Transport and Service Platforms, Telefónica: Complementing a bit. In Telefonica, we have different set of operator and there are some that are more capable than others. When we introduce disaggregation normally in any network level, we try to make it simple for our operational businesses. So we define the role of an integrator and having the responsibility of integrating the whole thing. So they perceive the new solution as a new vendor like any other. When this is already introduced, then we split the responsibility and we take in the operational businesses more and more responsibility for getting maximum value of the disaggregation. Because if not, you will handle like any other network element, it will not be useful this question.

Jean Louis Le Roux, Executive VP International Networks, Orange: So within Orange, we plan to operate ourselves with our teams and network. But this disaggregation is a great opportunity to drastically change the way we operate the network. Today we have teams, the things, the build, the run. It is well organized, well separated, et cetera. The idea with this evolution to our cloud-based networking, we really move to DevOps, BevNetOps, real DevOps in the core network. And this is a key enabler for this.

Yigal Elbaz, SVP Network/CTO, AT&T: I think what you’re hearing is that there’s two type of innovation there’s. One area of innovation is develop a product. The other innovation is the tech insertion. How do you take new capabilities and insert them into the network at scale? By the way, that’s much more complicated than develop a new product. How you do this at scale? How do you do that tech insertion. How do you make sure that you bring new values in? How you operate this? How you manage this? We take pride for being able to do this at scale in so many areas of our network over the last several years. But there’s a lot into that. And you ask a good question about how we pay attention to this.

Okay, I think we’ll end with that. Unless there is anyone eager just to be respect all of your time.
Thank you for all of you who came.