KubeCon + CloudNativeCon EU 2025 | Ranny Haiby, The Linux Foundation

Clips
More from KubeCon + CloudNativeCon EU 2025

Ranny Haiby

CTO Networking and Edge

The Linux Foundation

play_circle_outline Transforming Networking with Anuket and CNTI: Linux Foundation's Focus on Communication Service Providers and Enterprises

play_circle_outline Desegregation of software and hardware in network functions for scalability and uniformity.

play_circle_outline Edge projects like EdgeLake and InfiniEdge AI for running inference at the edge.

play_circle_outline Focus on AI applications for network operators for efficiency and customer care.

Info
Transcript

Ranny Haiby, The Linux Foundation

Ranny Haiby

CTO Networking and Edge The Linux Foundation

Exploring Networking Innovations with Ronny Haby at KubeCon+CloudNativeCon EU 2025

Join us as we delve into the forefront of networking technology with Ronny Haby, Chief Technology Officer of networking and edge at The Linux Foundation. At KubeCon+CloudNativeCon 2025 in London, Haby explores the intersection of Linux Foundation Networking projects and the Cloud Native Computing Foundation's initiatives.

In this insightful discussion, Haby highlights the convergence of IT operations and network functions within large enterprises and communica... Read more

explore Keep Exploring

What is Linux Foundation Networking and how does it relate to CNCF? add

What was the trend about a decade ago towards the desegregation of software and hardware in networking infrastructure? add

What is the potential impact of running inference at the edge on the widespread adoption of AI technology? add

What kind of applications would you like to see developed using AI for network operators, and how do you believe open source projects can help facilitate this development? add

bolt Powered by CUBE AI

Ranny Haiby, The Linux Foundation

search

Savannah Peterson

>> Good afternoon, open source fans, and welcome back to London, England. We're here at KubeCon and CloudNativeCon. My name's Savannah Peterson, as always at this show, joined by Rob Strechay. Rob, we're going to get to talk about something we haven't gotten to talk about a lot in this next segment.

Rob Strechay

>> No, I'm excited for it because I think, again, when I was at that hyperscaler, this was one of the topics that always perplexed me the most about how this would evolve and how networking would evolve in general.

Savannah Peterson

>> Yeah, I've been talking about it a lot just coming off of MWC, as well, and super excited to welcome Ranny Haiby to the show. Thank you so much for taking the time to hang out with us.

Ranny Haiby

>> Sure. Thanks for having me.

Savannah Peterson

>> So just to get everybody up to speed, give us the lay of the land and explain a bit about what Linux Foundation Networking is.

Ranny Haiby

>> Yeah, sure. So we are a sub-foundation of Linux Foundation, and we focus on hosting open source projects that are related to networking, as the name suggests, but especially the networking of communication service provider, telcos, and also very large enterprises like Walmart and others.

Rob Strechay

>> We were talking beforehand, and I think one of the things that would be good is helping people understand the intersection between what LF Networking is doing and what CNCF is doing. Because on the keynote today, there was a lot of talk about networking being done, and I sit there and I go, "Yeah, it's not done." I mean, there's still a lot to be done, especially when you get to the provider side of the networks.

Savannah Peterson

>> Yeah, wow. Interesting.

Ranny Haiby

>> Yeah, absolutely. So historically, if you think of a network service provider, they used to have two sides of the house. There was the IT or CISO side of the house, which looked very much like a large enterprise, and they've been adopting cloud-native and Kubernetes for a very long time. Then there was the network operation side of the house, which until about a decade ago consisted of dedicated boxes that were produced by companies like Ericsson, Nokia. And it was large bespoke boxes and you couldn't change the software. It was tied to the hardware. What happens about a decade ago is a trend towards desegregation of the software and the hardware where there were pretty much general-purpose compute hardware, and then the software of the network functions. We call them network functions, those are the routers, firewalls, whatnot, that connect the devices to the network. They were running as a software. Now, once that transition was complete, the next logical step was to see how we can take all the goodness that happens in the IT and enterprise networking and infrastructure and apply it to networking. So what's happening is kind of a blurring of the line, if you take a company like AT&T, between the CISO side of the house and the IT side of the house and the networking side, and they're trying to make it as uniform as possible so they can benefit from the economy of scale and use the same tools and methodologies on both sides. And what we're doing at the Linux Foundation Networking is we're building open source projects that help carry out this transformation.

Rob Strechay

>> Yeah, and I think just to build on that second, because again, with Jonathan Bryce being on OpenStack, coming into the Linux Foundation fold, I was a member of OpenStack community and helped build, for one of the people you mentioned, a lot of their virtual network functions out of OpenStack on a stack of our servers and purpose-built about literally a decade ago. So you're taking me back to my old home from that perspective.

Savannah Peterson

>> Are you getting a little nostalgic over here?

Rob Strechay

>> I am a little nostalgic about it. But when you start to look at that, how far has that come? Because now that that's coming into the Linux Foundation, as well, where does what LF Networking and what had been done in there... Because then you have OpenRAN and a whole number of other things going on, as well. Help people understand the landscape.

Ranny Haiby

>> Yeah. I understand how it's complicated to the observer from the outside, but that trend you mentioned, it was called NFV, network function virtualization, and OpenStack indeed played a large role there. So that trend evolved over time. We had a project under the Linux Foundation Networking called OPNFV, which was about the specifics of how you take OpenStack and apply it to those network functions. That project evolved over time, absorbed into a different project called Anuket, also under Linux Foundation Networking, that kind of tried to blend in all the transition to more cloud-native architecture, containers, Kubernetes. So that is still ongoing. So we have this Anuket project. We have another project called CNTI, the Cloud Native Telecom Initiative, that really helps communication service providers and the vendors to transform those network functions into something that can run on a Kubernetes infrastructure. So under that project, we have a set of best practices and test tools and certification that helps someone who works, say, for Nokia or Ericsson and builds those network function, adapt them to the world of Kubernetes and the Kubernetes ecosystem and successfully transform them. And that, again, is all towards that end goal of using this uniform infrastructure in both the IT and the network side. That is all based on Kubernetes and what we have here, all the ecosystem around Kubernetes.

Savannah Peterson

>> You're hitting on one of the themes of the show, I feel like. And I mean, we know this a bit in our worlds in general, but with standardization comes simplification. You have that uniform platform, you have a lot of different experiences you're trying to gather. You're also getting a lot of information and data out of silos. You mentioned the companies that you're working with are everything from telcos to Walmart. We're talking about an incredible amount of data, and probably some very eager community members who are trying to come up with solutions. Is the best practices similar for both of these types of two very different companies, or is there a little bit of difference?

Ranny Haiby

>> Yeah, it's a good question. I think you're correctly stating that everybody's interested in data and leveraging data for AI and getting more efficiencies with the AI and large models and the latest and greatest from that world. I think, if you look at an enterprise, it's a more common problem. But when you go into the telco service provider, then there are some challenges because their data is quite unique. So the data coming from the network, from the RAN, the radio access network, contains all sort of personally identifiable information, PII, that is unique to telco. So when you try to apply data anonymization general-purpose techniques to that, they don't identify those fields that are very unique to the telco world. So we've actually launched a new project this week with the help of Infosys that donated the seed code called Essadum. And Essadum is going to help all our member companies, but anyone interested because it's an open source project, to take the data, prepare it, and build AI applications on top of it. And this data anonymization I mentioned will be one piece of it, and it will be tailored for these networking telco use cases. So there are many very good general-purpose framework for building AI applications, but this one is going to focus on the needs of the communication service providers and large enterprises and allow all of them to create AI applications efficiently and commonly develop the common parts of those applications so they have the resources to invest in the top layer where they can differentiate and bring the real value and not waste time on reinventing the wheel on the bottom layers, which tend to be very common in any AI application you're building for networking.

Rob Strechay

>> Yeah.

Savannah Peterson

>> I can imagine there's some people very excited for you to build that out and come to that conclusion and solution.

Ranny Haiby

>> Yeah, we're already seeing a lot of interest because it hits the spot for many organizations that are dealing with, "Okay, how do I develop AI applications?" So all of them want to hit the ground running and not invest a lot of time in preparation of the data and then building the framework. So I think once this project is rolled out, and once we have more players, I think many companies will be able to benefit from that.

Rob Strechay

>> One of the things we've been talking about and we've been seeing is really the explosion of small language models and the fact that they're going to live at the edge. It's inference at the edge and you're going to have a lot... Is that a lot of what you're seeing from the constituents of LF Networking, they're like, "Hey, we need to get ahead of this because we've got this wave of AI that's going to move out to our towers, out to the edge, out closer to the consumers of that AI"?

Ranny Haiby

>> Yeah, absolutely. So my full title is CTO of Networking and Edge, so I deal with some of our Edge open source projects. And yeah, I think AI is boosting this concept of running inference at the edge. We have a few, again, recently announced projects around that coming from LF Edge. One is EdgeLake, which deals with the data at the edge and having distributed sets of data without having to transfer the actual data from Edge to cloud. And then another project named InfiniEdge AI that, again, brings sort of many of the building blocks that are required to run inference at the edge as an open source project. So yeah, we definitely see that trend, and many organizations, both communication service providers who have the mobile edge, but also other edge providers or enterprises and industrial IoT companies that have edge infrastructure, they want to run inference at the edge. So yeah, that is a use case that we're seeing a lot recently.

Savannah Peterson

>> I think that inference at the edge is what's going to make AI real for the majority of the world, beyond the browser. We've obviously got the way that we're interacting with it now, but I think what we're talking about is so powerful and so interesting. I'm wondering if you have any, and I'm not asking you to pick a favorite child, but if you have any use cases or examples that you've seen that you can share from the community that illustrate best practices or very exciting case studies around that?

Ranny Haiby

>> Yeah, I mean, I don't want to limit the imagination of the viewers, but I think that what we're seeing some use cases are related to autonomous vehicles or vehicle-to-vehicle, more specifically.

Savannah Peterson

>> Yeah.

Ranny Haiby

>> So we have one case coming from one of our member companies where they have vehicle-to-vehicle data flowing into edge and running inference at the edge to take... Because it really matters when you need a quick response, like you want to avoid collisions or have the vehicle communicate with the infrastructure for safety. So those kind of safety use cases in these cases makes a lot of sense to run inference at the edge and have a very quick response time for these safety use cases. But there are many other use cases, so I think we'll see more and more coming.

Rob Strechay

>> So one of the things you hit on a little bit, I want to kind of dig in a little here is because when I was doing the NFV stuff back in the day, one of the big things of moving from SS7 to NFV was really around security and how do you get security right? How does LF Networking and edge really take into... You talked about the PII stuff, but how does security fit into all of this?

Ranny Haiby

>> Yeah. When you deploy your project in a communication service provider environment, you can't ignore security. So I think the LF Networking projects have been in the forefront of adopting security standards for open source for many years now. We are kind of relying and adopting on the work of another LF initiative called OpenSSF, and they create best practices and tools for open source projects to become more secure. So there is a saying that just because the code is open source and there are many eyeballs, it automatically becomes secure. It's true to some extent, but you need to do real work to secure your software, and we secured a supply chain of how you build the software and make sure there are no loopholes or backdoors on the way. So many of our projects are using tools from OpenSSF to create things like software bill of materials, so that's embedded in the CICD pipeline of the project. So when some of our projects deliver the open source software, it's also delivered with a software bill of materials that lets you know what's in the mix and that you're not vulnerable to some vulnerabilities. We also have standard security checks that run continuously, again, as part of the CICD, and we have dashboards provided by the Linux Foundation that show where we stand, which are the CVEs. All of our projects also have procedures to deal with security reports and CVEs. So each of these project communities have designated people who are actually there to respond to provide the patches and make sure that we address these vulnerabilities. So yeah, security is top of mind for many of our projects. It's always hard to make developers think security first, but I think over the years we've made it part of our process of building the open source software, so they cannot ignore it. Even if they didn't think about it to begin with, they are being constantly reminded that they need to make sure we produce secure software.

Savannah Peterson

>> Well, and that's how things scale, and that's how you end up getting integrated into these larger companies. It's got to be secure.

Ranny Haiby

>> Yeah.

Savannah Peterson

>> I mean, it's absolutely table stakes. Ranny, I've got one last question for you. When we are hanging out in Amsterdam a year from now at KubeCon, what do you hope to be able to say then that you can't yet say today?

Ranny Haiby

>> Oh, I think I want to see some of this AI framework that we're building really coming to fruition in the form of applications. So I want to see many more AI-driven applications for network. I mean, many of our member companies talk about AI-native networks, but what does it even mean? So I would like to see, with the help of our open source projects, a lot of AI applications for network operators, whether it's on the network operation side, whether it's the customer care and whatnot, whether it's efficiency. But I really want to see the power of AI and machine learning being put to use for a good cause and really make the life of a network service provider easier and the life of us as the customers more exciting and more services coming out with the use of AI.

Savannah Peterson

>> Excellent. Well, we certainly look forward to talking to you about that next year. Thank you. This must be a very busy week for you. We really appreciate the time.

Ranny Haiby

>> Absolutely. Thank you very much for having me.

Savannah Peterson

>> And thank you, Rob, as always.

Rob Strechay

>> Oh, yeah.

Savannah Peterson

>> And thank all of you for tuning in. We're here in London, England at KubeCon and CloudNativeCon. My name's Savannah Peterson. You're watching theCUBE, the leading source for enterprise tech news.