 Welcome to the CUBE's coverage of KubeCon EU 2024, live from Paris, France. Join hosts Savannah Peterson, Dustin Kirkland, and Rob Stratche, as they interview some of the brightest minds in cloud native computing. Coverage of KubeCon cloud native con is brought to you by Red Hat, CNCF, and its ecosystem partners. The CUBE's coverage of KubeCon EU 2024 begins right now. Good morning, cloud community, and welcome back to spectacular Paris, France. We're here at KubeCon cloud native con, CNCF's flagship European event. My name is Savannah Peterson. Delighted to be joined by my fellow analyst Rob Stratche. Rob, did you have any pastries this morning? I did not. I did have a baguette, though. You had a baguette this morning? I had a baguette this morning. That was my breakfast, so. I'm actually a little jealous. I'm not allowed to. It was fantastic. You didn't want to bring a baguette for the rest of us? It was on the walk. Oh, it was on the walk. All right, all right, all right. Well, at least I'm virtually having a baguette right now, because we're talking about virtualization on this next panel. Very excited to welcome our two guests, Alice and Fabian. Thank you so much for being here. How y'all doing? All good? Yeah, I'm glad to be here. Do you like, is this show a big event for you? Yeah, sure. You've gone, it's like the event. It is the event. Yeah, we're also really big, super deep nerd fans. Fabian, you were on stage today. Yeah, indeed. Tell us about that. Tell us about that. Tell us about that. You're on this stage now, and we want the highlights. It was, you're like, I feel, you know, it's great, right? To have that opportunity and to have like such a great co-speaker who can't be here right now, but like, it's amazing, right? So you have somebody, they've adopted Kuber, it solves a problem to them, and then you can get into such a stage. I mean, it's like, wonderful. Well, it was Goldman Sachs who was on, talking about how they're using Kuvert to actually move virtualized machines, or virtual machines into their Kubernetes environment, which I think is, I kind of talked about it yesterday, how I was surprised we hadn't talked about it, and I totally forgot you guys were coming on today, but now we get to talk about it. Or did you forget, subliminally, I think you were definitely teasing this interview the whole time. I may have been teasing this interview the whole time, but I think, again, it's such an important, topic, I think, and a project in general, because of the changes that are going on in the industry, with Broadcom buying VMware, there's a lot of service providers, for instance, that are looking for other options, for doing hosting of virtual machines, and things of that nature, because of the contract changes that have gone on underneath there. Do you see a surge of people being interested in Kuvert since all of this has started happening? Sure, I mean, even here at KubeCon, we got many questions, how people can transition from VMware to Kuvert, if Kuvert is stable enough to be using production, so we have seen a lot of interest in the last month. So. I know, I bet you're not surprised by that. I'm not, I'm not. So, what's new with Kuvert? What's been going on? So, right now, Kuvert is like reaching kind of maturity, so we are focusing more in cleaning up the code, get more people involved. So, there has been many initiatives to improve the code base, to clarify more the ownership. There has been also a lot of improvement in the user guide. Many people are joining the community, this is of course the first entry point for the users. As Alit just said, right, so we're working a lot, like, you know, Kuvert, with Kuvert we had a race, right? I mean, we, you know, with Kuvert we want to provide, we are providing a virtualization solution based on Kubernetes. And just one addition to what you said about like Goldman, like, you know, important is, you know, Kubernetes is becoming the infrastructure and VMs become one workload that's being run there, right? And Kuvert is enabling to do that. So we've been working on Kuvert right to enable all these features that you expect from, you know, these virtualization systems that are out there today. And we've achieved much of that over the past few years, like some big things like hot plug, perceived memory resources are still, you know, they need to GA, memory over commit, higher density, that's something we're working on. But we're switching to, you know, being even more stable, right? We provide seamless upgrades, but we want to be more stable. We're working on, you know, making it easy to maintain, because if you switch to Kuvert, you can rely on it, right? And that is, that's what we want and have to live up to. So we're changing focus a little bit, while not, I mean, while continuing to focus on innovation as well, right? You know, new features have to go in, but the foundation needs to be solid. Yeah, another interesting thing is, usually, virtual machine are hard. They have many options. So we have tried to simplify the deployment. So there is a new feature. Super important, everyone wants to decrease complexity. Yeah, so for example, we got so many questions about Windows, how to deploy Windows VMs, so you can define actually, if you want to run a Windows VM, there is like something called Instant Types, and this helps users to deploy a virtual machine, so this kind of template that will set all the options for you. So this is, again, to make it easier for users to deploy VMs. Everyone wants that easy button. We talk about it a lot, and especially when it comes to Kubernetes, especially when we're talking about AI and everything else. You know, you and I were talking about your daughter's senioritis earlier, and thinking about graduation. Where's the project maturity right now? Fabian, I'm going to go to you on that one. Yeah, so, you know, I think we joined the CNC of incubator, Sandbox in 2019, incubation 2022. 2022, yeah. And that was always a good reflection of the project's maturity, I would say. So I think it's just, you know, the natural next step to now work on graduation. It totally aligns to what we're doing, like on a day-to-day basis, like working stability, finishing the, as I just said before, right? So yes, graduation is definitely something we're working on this year. We're working actively to reach out to, to end users to more platform providers, like we have a lot of platform providers who are using Kubernetes and offering Kubernetes on top as a service or, you know, using it by themselves. So we're working with them to simply complete that whole graduation. It includes security reviews, involves governance, and involves end users and other adopters. How many contributors are on the project? We have 150 contributors. You bring back the days for you, you know, than others. And I'm even more about the companies because the company ratio increased over time and it's always good for a project, right? To not rest on a few companies. So it's great to see that more companies become interested, regularly contribute to Kuvert. So that's just great to see. Do you see people can, what's, I guess, the ratio between people who are vendors contributing versus non-vendors that are contributing to this, do you think? I would like to take that offline and work with you to digest the data. That sounds good. I think, like, you know, Kuvert, Kuvert is for sure very interesting to platform providers. So vendors, what you could say, and they are for sure, I would say they are more dominant, right? And now we're going to be... But some end users are also contributing, right? And where are kind of the requirements coming from? Is it people saying, hey, this is what we've seen in KVM, this is what we've seen in VMWare, these are the things that are missing. How is that working with the community? So, I mean, of course, people coming from traditional virtualization plus what they expect, certain feature to work. Like, for example, hotplug as Fabian mentioned, but we are also getting requirement like from container perspective. Like, I don't know, like setting certain environmental variables or passing certain kind of files. So, I would say it's kind of a mixture, but yeah, a big part of the requirement is for sure coming from virtualization, the virtualization. And it's leveraging, one of the other keynotes, this morning they were talking about the storage aspect and it's leveraging CSI to be the connection back to storage. And is that, are you being able to leverage a lot of the different things that have been done for more ephemeral container types, but for something that needs to be more persistent? Yes, definitely. So, Kuber doesn't make any difference between storage provider. So, if you have a CSI plugin, you can work with file system and you can also work with block devices. So, yeah, generally, if you have your CSI plugin, it can perfectly integrate with Kuber. And can I just add something? I think that's true for so many aspects, right? I mean, why did we build Kuber and Kubernetes at all, right? Because Kubernetes is providing so much, right? From an infrastructure perspective, CSI, CNI, but also all these other tools, like patterns for observability, patterns for our back, right? Policy enforcement, such that stuff. That's why we built Kuber on top of Kubernetes to leverage it. Why should we reinvent it again for VMs? How can we build such a wonderful ecosystem that we have with the CNCF ecosystem? Yeah. We could not build that for VMs only, right? So, like by embedding VMs into Kubernetes, we can simply benefit from that ecosystem. That's providing so much, it's like. I mean, it makes sense. Yes. Everything makes sense. So, what's next? What are you going to build next? What are the contributors going to make next? What sort of features can we expect? What's on your roadmap for this year? Yeah, exactly. Yeah, give us the preview. So, live update of resources is of course a big topic. So, we just added CPU and memory hot log. So, we were already supporting storage and network interfaces. So, we can expect also in the next future to have additional features in this direction. Do you see different workloads coming to Kubevert that are non-traditional VM workloads, like some of the AI stuff, some of the database stuff would make sense to me, which is not a, I know you can do persistent databases on, and I sat in a Postgres persistent database talk, provide data on Kubernetes Day on Tuesday. Do you see those types of workloads really being driven to Kubevert as well? Go for it. I mean, we really see user using GPUs, so we can expect that they want to run AI workloads. I don't think we have many feedback if which kind of workload runs inside VMs, but yeah, the interest in GPUs is of course there. Yeah, like the end user feedback, like sometimes it's difficult, but I know for sure that like, this is finding its way into Kubevert, right? I mean, in the beginning, there was like uncertainty, how does it behave, but KVM is old, right? I mean, it's been used in OpenStack, it's been used in other projects out there for years. So what I want to say is, yes, we see AI mail workloads running in VMs because there's some benefits to it as well. We see tech workstation use cases being reduced, databases, everything, and you know, the nice thing is it's run as before, right? So you get the same stability, the same compatibility that you, you know, have another KVM based stacks. It's just that you manage it differently and we see that. I mean, just to conclude that like, with all the preferences and instance types that we've been doing, like Windows, for example, is like a common workload in VMs. And it's being used. We have complaints about not being able to automate as much, which is difficult with some operating systems, but that's a good sign, right? Because we see that those workloads get into Kubevert. Another interesting use case is deploying Kubernetes in Kubernetes. So Kubevert is like the deployment where you have maybe additional cluster, nested cluster. So this is also another perfect use case for virtualization. Yeah, I think it's really interesting. Actually, I want to dive into the AI side of things. What sort of trends are you seeing in terms of how people are using Kubevert to manage those workloads? AI will rule the world. Yeah. And... Enough said, soundbites sorted. Kubevert can be decided, right? I think AI is finding its way into, you know, into an operator's daily life. Like, I mean, AI is approaching us from all sides, right? It's helping operators. It's helping us with data, big data, right? To analyze it. All the large language models, they're about big data. But we also have it as a workload. So I think, like, we will see AI, you know, approaching Kubevert from many sides as a workload, but it's from the Kubernetes side. Like, everything that's like AI and integrating with Kubernetes, we're expecting it, you know, at least touch Kubevert as well to some extent. But we need to see. I'm not aware of anything specific related to Kubevert in AI at this moment in time. Yeah. Do you have any thoughts? I think, as I said, then use of GPUs, really I think it's the result of the interest in AI workload. And I mean, NVIDIA is one of the main contributors to Kubevert, so definitely people are interested in using AI workload with VMs. I was going to say, they were on stage yesterday talking about the APIs and things of that nature. So is that a big direct input into how Kubevert is using the GPUs and the hardware underneath? So a big shout out to them. I mean, they're really taking, it's really serious. They do a lot about scale and performance in Kubevert. Yeah, they have a zig scale, right? They're driving it, that's great. And yeah, they optimize it for their use case, including GPUs. Yeah. How are things go from Kubevert to being in OpenShift virtualization? And what are the benefits of that? Because I mean, again, you are here from Red Hat, so you can put your Red Hat on for a second and talk about that. How should people think about that and how close is it to Kubevert? So I mean, usually installing and managing Kubernetes is hard. So you can really think Kubernetes as Linux and the same Red Hat to OpenShift. So OpenShift really gives you the entire package and the installation. And we work very closely with the OpenShift team, so it's very well integrated. So when you do enterprise, you probably want to have a food solution. And this is the benefit that you got with OpenShift and Kubevert and cloud native virtualization. So it's just part of the whole package? Yeah, you can install it through the operator hub and it works like every add-ons from OpenShift. Yeah, like, first it's very close, right? I mean, for sure, like, Kubevert is built. Like, I mean, we build it as Red and chip it with OpenShift. And it's tied into the user system because the reality is Kubernetes alone and Kubevert alone, you seldomly see it, right? But it's with any other project, right? You need that ecosystem, right? You need observability, you need storage, you need networking. And OpenShift, after all, OpenShift by itself is an enterprise Kubernetes platform that is giving you, like, this boilerplate, right? To focus on what is bringing value to the user, right? About running the applications. And OpenShift virtualization is just benefiting from this, right? We have OpenShift as a solid platform and we add virtual machine support as part of OpenShift virtualization and it ties naturally into what the admin is already doing, right? So they don't need to re-skill it, so they can use what they know, working with Prometheus, right? Working with Loki. There's nothing they need to learn, new excitement, except for working now with the VMs as well. But that's just another workload, nothing. Just another workload. Just another workload. Love the whole ethos. You've had some great soundbites, both of you today. Last question for you as we wrap up. What do you hope you can say at the next KubeCon that you can't say today? Ooh, good questions. The next KubeCon that we can't say today. Yes. Where do you hope we're progressing? Or the project's progressing, or? So definitely getting more contributors, that will be a great achievement. I like that. I think seeing that I would, now, switching forward to 2021, it's great to see that finally people realize that Kubernetes can be, it's a compute platform that can be run on bare metal as well. It's not only for application developers. It can also be a tool, and that's fantastic, being on bare metal, used to automate your work. I mean, containers are not just a tool to drive applications. We need to understand, as an administrator, I need to administer my infrastructure, and the term operator is coming from the fact I'm codifying operational knowledge, right? So why can't we codify an administrator's knowledge to operate our infrastructure? So I would love to, I'm expecting, no. I'm looking forward to see that Kubernetes is recognized as such, right? It's a compute platform, right? And that is running multiple workloads, so as a tool for administrators as well. I love it. That was great. You're both absolutely fantastic. Thank you for joining us this morning, making the time and your busy schedules between keynotes and everything else. Rob, thank you, as always, for your wonderful insights and questions. And thank all of you for tuning in to our three days of live coverage here at KubeCon Cloud DatavCon, CNCF's biggest European event. My name's Savannah Peterson, here in Paris, France. You're watching theCUBE, the leading source for enterprise tech news.