Exploring VMware Cloud Foundation and Kubernetes Integration with Broadcom’s Paul Turner

Hi, everyone. Welcome back here on Textron tv. If you're watching this on, on our techron uh, tv, uh, feed, great.

If you're watching it on demand or somewhere else. Great. Thanks for joining.

I'm Alan Shimel from Techstrong, and I've been looking forward to handling this interview for some time. I want to introduce you to Paul Turner. Paul is VP of Products for the VCF division of Broadcom.

Hey Paul. How are you, man? Hey, I'm doing great today.

Yeah, it's a, a wonderful Thursday here. So What, what could be bad? You're out in the vineyard probably having a nice glass of I wish day, cover Day.

I wish I was out in those vineyards. So those vineyards are actually, I, I like, I love their Sonoma vineyards. I love biking up there.

It's a phenomenal place. So anytime anyone's around there, go biking, get up, get out in the fresh air, enjoy the vineyards, enjoy the wines. Absolutely.

Paul, I mentioned you're VP of VP of products of the VCF division. Uh, most of our audience, look, they know who Broadcom is, they know VMware, but there might be some people who don't. Talk to me a little bit about what, what VCF stands for and what the VCF division of Broadcom's about.

Yeah. So if you think about, uh, you know, VMware Cloud Foundation and, uh, the main focus, you kind of think back to VMware, right? Providence of us 25 years ago.

What did we do? We didn't just increase, you know, deliver the best virtualization possible to actually get efficiency off servers. We changed the way data center operations worked because data center operations, you could suddenly actually define kind of perfect specs of what your applications, your VMs became, your encapsulation of your applications and your services.

You're able to deliver them onto this IT infrastructure and do it much more efficiently and get density, get cost efficiency. But you look at it today, while that's great server virtualization to actually deliver an agile cloud environment like you can do in the public cloud, you need to deliver more. You need to virtualize your data center.

VMware Cloud Foundation is about virtualizing your data center, your infrastructure. That means your compute, your network, your storage, your automation you need. So you can deliver for any application containers, VMs, you can deliver an application platform surface that they need.

And it's very easy to operate for the big enterprises, or you can deploy it up on a public cloud. We're we're supported on all of the major public clouds and the set of our service providers. So, so think of it as taking virtualization to the next step.

It's about now delivering a full cloud service to the customer. So you get that agility of operations and that delivery the customers need. Absolutely.

And of course, right. I always like to tell people VMware may not have invented the hypervisor, but they certainly are the people who brought the hypervisor to the market in light, in ways unseen before and, and really was, it was the, the spark that ignited the whole cloud movement. Yeah.

Yeah. Innovation is all, Whether private public Innovation's all about bootstrapping on top of each other. How do we keep continuing to innovate and build better products?

You know, what we are doing now with Cloud Foundation AWS showed us that we could actually build on top of virtualization and deliver a cloud service. What we're delivering in cloud foundation is really that full encapsulation of a software-defined version of that. So everyone bootstraps.

Yeah. Abso well, Fred, Brad Felds a big vc. He always tells me 99% of what we do in technology is evolutionary, not revolutionary.

Yeah. It's that 1% revolutionary that we all build on top of and bootstrap. Yeah.

To your words. Hey, Paul, just real quick before, give him a sense of your, how long you been with VMware? What else have you done?

Yeah. Yeah. So about eight years actually at VMware.

And, uh, I actually led, uh, the first integration actually of Kubernetes into VMware and into vSphere, uh, back in version seven. Uh, and so it was, yeah, it's kind of exciting time. It's nice to see us actually take that much further.

Uh, we've done a lot of work in it recently. Uh, my providence, uh, i storage background, uh, big, you know, originally a software developer out of Oracle doing cluster development, uh, way back in the mid nineties. Uh, that kind of stuff went on into, you know, product management and leadership roles across the, across the industry.

But, uh, yeah, glad to, glad to be here. It's nice to, you know, know that, you know, the impact that you can make as VMware. We've, you know, nearly 400,000 customers and how do we actually deliver cloud platform service to them?

It's, it's exciting to be here. Absolutely. So, Paul, you mentioned, you know, involvement with Kubernetes.

Let me tackle this one head on for whatever reason, and I don't know why or how, but there was this perception, or there is this perception, maybe it's ongoing mm-hmm. That when you are, you know, we talk about modernizing applications, hosting applications, you have a choice. We can go the virtualization hypervisor VMware for all intents and purposes.

We could go the VMware route or we wanna go the, you know, the new kid on the block with cloud native, you know, microservice, Kubernetes kind of stuff. But for some reason, the perception, as I say, maybe it's a misperception, was you could do one or the other. You, you can't have peanut butter with your chocolate.

Now, people in the know out here, our audience, they've been hosting Kubernetes on hypervisors from day one. So I don't know where, how this whole thing blew up. But give us, first of all, if you have an answer to that, I'd love to hear your theory.

But secondly, let's talk, let's blow that theory out of the sky here. Yeah. And, and talk about Often, often, uh, often vendors lead to our biases.

Uh, and sometimes vi vendors are biased in their view in terms of whether I should drive it on bare metal that may suit certain vendors. At the end of the day, operationally, it's a heck of a lot easier to run it on top of virtualization. And when you look at, even when you run up on the cloud and you're running, you know, A-W-S-E-K-S, uh, GKE, you're actually running on virtualized environments.

Most is running on virtualized environments. Red Hat, OpenShift, you know, probably 70% of their deployments are running on top of virtualized environments. It, the pure nature of it is, it's actually very beneficial to combine the two.

You get the agility of Kubernetes, the kind of desired state way of managing infrastructure, kind of GI ops led operation. Uh, it's a very, very nice way to actually deliver an application service and to, and for DevOps teams to be able to do what they do, you know, deliver applications. But the systems teams still need to patch, maintain, update, drivers, firmware, all of this kind of stuff means they've gotta do maintenance on servers.

And that's what virtualization really helps with. Totally. Non-disruptive, your applications and your application surface.

You can manage infrastructure, you can increase it, you can scale it, you can dynamically manage it. That's the difference. And, and of course, with Cloud Foundation, you, you, you know, Kubernetes doesn't just require, you know, virtualization of the compute server.

It requires virtualization of your network. 'cause you actually have to do your CNI adapters, your CNS adapters. You have to integrate your storage in.

That means it's all software defined storage that you actually need. And so it, it's a, it's a kind of, the virtualization benefit is significant compliment, and it's actually a necessary compliment, uh, for running Kubernetes. And it's the same as you would do up on AWS, it's the same as you would do up on Google with GKE.

It's the same as you do with VCF. And our integrated service, which is called VKS, uh, uh, the VMware Kubernetes service. You know, the thing I've learned in my 30 plus years, nice to have.

It's not enough. Must have, Must have is what's Right. And there are some must haves of running Kubernetes.

I, there, there, you're right. There's a percentage of the market that wants to run Kube on bare metal. Mm-hmm.

And they, they have their reasons. But that's really a small minority of the market here. Overwhelmingly, I think the market has voted, you know, with their pocketbook and their feet and eyes and fingers that you run, you run Kubernetes in a virtualized environment.

Now, as you mentioned, you can run it on Amazon, on anyone, you know, they have several platforms, the rest of them do. But let's talk, Paul, if you will. What are the distinct advantages, the kind of killer advantages that you get from running Kube in a, in, in a VCF in a VMware environment?

Well, let me, let me touch just a couple of things. So, so, uh, one of them is, is think about running a Kubernetes environment. I have to actually deliver Kubernetes.

Kubernetes has this wonderfully short, uh, uh, support time window for Kubernetes once every, what is it? 18 months, 16 months, somewhere around there. So you actually naturally need multiple versions of Kubernetes.

So, so we provide multiple versions of Kubernetes, but you don't just need a Kubernetes runtime. You need a, a registry service like Harbor. You need a service mesh.

'cause if you wanna actually do function as a service and integrate kind of, uh, applications together to actually deliver a container application, then you actually need a service mesh like Istio. You need, uh, an identity service, a pin AED integration. You need Prometheus for monitoring.

You need Valero for backup service. You need your network integration, CNS and your CNS your, your cloud native in, uh, integration for networking, your CNI or your CNS you of many customers want to actually manage also VMs, uh, and deploy them. All of these services you have to be able to deploy with VCF, we give 'em all outta bucks dead easy.

We manage all of the long-term support for the Kubernetes platform. We actually manage the delivery of those platform. We run them as services.

So for a DevOps team, we take away the pain. Right? If you think about a, a, there was A-C-N-C-F survey, which I, which I refer to, and I, I thought it was a very interesting survey.

It said 46% of the people, and it was actually 2024 survey. I know they just updated their 2025. But 46% of the people running Kubernetes say the biggest challenge is complexity.

Biggest challenge is complexity. And, and, and we as an, an industry should be abhorred by that. We should look at that and say, that's totally wrong.

It's not the way the world should be. I've been yelling about it since the first time I saw it. I came back and said, we'll, never catch on too hard.

And the people who are fixing the complexity are the most expensive resources. They're the developers. The Developers, The developers who are actually now having to maintain versions of different components.

Like I mentioned, pin, aed, Prometheus, Valero Harbor, Istio, you know, you, you go on, all of these services should just be there. You, you gotta simplify it. You gotta, and, and they become part of the platform, and the platform is just there for you.

And, and that's what's so exciting by the work we did. We delivered the first Kubernetes, uh, in vSphere, uh, way back in version seven, but delivering a full Kubernetes platform runtime with all the services, with the long-term support, that's what's in our BCF nine. And that's what's very exciting.

Yep. So, couple things. First of all, you just made the argument why we need platform engineering.

Yes. Right? Asking the developers to develop their own platform Is really, it just doesn't make sense.

It's wasteful. Yep. Right?

And hence, you, you need platform engineers to put these platforms in, in place. Now look, not everyone is gonna be as familiar with the, uh, VMware nomenclature numbers and all of that. The fact of the matter is VMware has been working with Kubernetes since like 2018 or thereabouts, if I'm not mistaken.

Right. I, I mentioned the first time I saw Kubernetes, probably 2016 or 17. I, I thought it would never catch on, which is why I'm still working and not out in Sonoma.

But, um, it was hard. It's been hard. It's always been hard.

You guys have been working with it since 2018. Yep. Talk to us how that evolution right.

Has evolved over the, from 2018 to today, and you know, where it was then, where it is now. You were just a cube con. You're up on the latest.

Let's hear what, what you've got. Yeah, Yeah. Well, you think about it, right?

2018, I mean, that was the very beginning of Kubernetes. So the, the founders at Kubernetes actually at Google, um, spun off, started a company. Heptio Heptio actually was acquired by the mware.

Uh, that led to our first integrated Kubernetes, uh, that we actually delivered into vSphere, as I say. Um, and in, in actually a version eight of vSphere. Uh, we actually made it free for everyone.

We included it in it's no extra cost, it's just part of the platform, version nine. It's also part of the platform. So it's part of that VCF platform.

Uh, so, so yeah, it's, it's quite interesting. And then you look at kind of some of the, the open source projects, right? You look at Andria on the networking side, you look at the Envoy service, you look at Harbor, the registry service, these are kind of core capabilities.

They were all projects led and delivered by VMware. So VMware is, is still today one of the most active contributors into, uh, the cloud Native Compute FO foundation, and into the open source of Kubernetes. Uh, we're a gold sponsored member, of course, of CNCF.

So, you know, we are, we are here to stay. We are here to deliver the best delivered platform for runtime for Kubernetes. Um, and to make it make it easy, honestly make it easy.

And the last thing that I think we can do that's kind of unique is one of the most traditional things that you see in vSphere, which is vMotion. Uh, vMotion is still absolute amazing. Uh, it's, it's a kind of cool thing, being able to non-disruptively move workloads around.

And you, you'd say in Kubernetes, that doesn't really matter, but Kubernetes, we're running on Kubernetes now. Big AI applications, right? Gen AI applications.

You're running a 70 billion parameter LAMA two model. Imagine you could take that LAMA two model that's running your PyTorch application, doing your inferencing, and you can magically move that when you're actually doing maintenance mode in a cluster, magically move it without any disruption to the application. That's what vMotion is all about.

You can do it on top of cloud native infrastructure, which is basically what we deliver with VCF. So it's, you're full of Kubernetes runtime, running your most modern applications, your Gen AI applications. You get all the benefits, of course, of the platform, which gives you all the services, but you get the benefits of the platform and cool features like vMotion, which allows you to do the magic.

Um, so that's, that's what, that's what's made, that's why there's not another virtualization vendor that's anywhere near as big with us. We do cool stuff that, that extend virtualization. Absolutely.

Lemme bring up another kinda thing that kind of gets me going. We live in a very complex world today, right? It would be, you know, if we just kept all our stuff in one data center, or even if we just kept all our stuff on one cloud provider mm-hmm.

Life would be so much simpler. But it's not. We, we, we, we, we have multi-cloud, we have hybrid cloud.

We're a little, we're a little in the data center, a little in the cloud, a little here, a little there. We, we, you know, and every cloud provider has their own hypervisor advisor that don't necessarily play nice with each other. Um, some people want Red Hat there or this one there.

And, and the, the people I talk to say enough, we want, I gotta be able to manage it all. I have to be able to manage it all. I can't manage 12 different platforms.

Mm-hmm. To me, this is, this is what VMware brings to the table. Yeah.

Right? Let's, you are talking about taking the complexity out of things. This, this is the real con complexity.

Yeah. Talk to me how, how you helped there. Yeah.

Yeah. Well, well, first off, I think customers will wanna run across multiple clouds. Um, Yes, they will.

They, they absolutely do. I think even in the data center, right? I'm, I'm gonna have, am I gonna bet on just one server platform?

No, I'm gonna have two. I want to have arbitrage between my vendors and you're gonna have arbitrage between your cloud vendors. So, but the problem is, is yeah, if, if you've gone and built the, where the platform service and the platform experience is different, and it's, but the problem is, it's, it's a different platform experience now between the different cloud vendors.

That's not good for you. That's why you're seeing us look at the infrastructure and say, we can build a common VCF infrastructure that can run in any cloud. We actually support it up on Google, support it on Amazon, we support it on Microsoft Azure.

We support it on-prem, same operating experience. You get complete motion of your applications and services that you need. Same operating experience.

Uh, and I think customers love that. So that's, that's one part of it. But the other side of it is you've got to then get behind Kubernetes open source and allow the developer and the application platform to be fully CMCF compliant, fully portable, so that it can work across any platform.

If you wanna move your applications and services and run them up in GKE native versus on top of our platform as, uh, uh, and, uh, VKS native, uh, you can do that. Uh, that's very important. And for us, we are committed to open source and maintaining complete compliant open source.

And we hope all vendors do the same. Because we saw what happened with Java, we ended up with this kind of mishmash, and everybody had to go and build applications per platform because it didn't give us the platform independence that we need. We can't let that happen Kubernetes.

It must remain open source compliant and cloud native. And, uh, and then we can allow the portability, the application there. 'cause that's where the main complexity, but they're two different layers.

How do I manage the infrastructure, how policies and governance rules and IT rules. That's what we do with VCF. So we get it everywhere, but maintaining the application developer experience, that's why Kubernetes is so important to us.

It's why we built it into the platform. Hey, we've gone on here 20 minutes or more. We haven't really talked about ai.

Yeah. You mentioned it kind of in passing, but AI's too important today to mention in passing. Mm-hmm.

What's happening with VCF and ai? How is it, how is that affecting what's coming down? What you guys are doing?

Yeah. What the market is wanting. So It's kind of interesting, you know, again, I'll, I'll split things into infrastructure first and then into actually the developer kind of experience for ai.

But, but if you think about the infrastructure for ai, we're kind of like where CPUs were, uh, 20 years ago. We are running utilization on GPUs, which are super expensive, but GPUs, server eight GPUs in a server loaded up, you're going to spend, you know, it's H 200, nvidia, you know, latest, latest kind of processors. It's gonna cost you about 300, $400,000 for that server.

And those servers are running utilization rates somewhere in the 20 to 22% range. That's back in the days when we used to do physical dedicated servers per application. Mm-hmm.

Right? It's a, it's begs for virtualization. And, uh, so that's one of the core things is why we've done all our work actually with Nvidia on something called VGPU.

And with version nine, we actually just have done the testing on DGPU testing with vMotion, like I said, being able to non-disruptively move workloads, 344% improvement in performance. Totally non-disruptively. Wow.

Yeah. You can run wow GPU as a service. So now I can virtualize, now I can start moving applications around.

I can look at balancing so that I'm optimizing the amount of memory usage that's on that GPUI can move workloads if I find that I can't, you know, cookie fit them into the GPU memory properly, I can move 'em and rebalance them elsewhere on that cluster so I can truly run GPU as a service. So that's what we do in the core is, is we think the whole market's gonna move to, it must be virtualized, um, because it's what happened in the server market. The same will happen in the AI market.

The other side to it though, is the application ecosystem. And a, and a big part of it, like all AI applications are actually built as container ready applications. That's the way we deliver application services.

We've worked very closely with NVIDIA on, uh, what we call private AI foundation, and that means all of the Nvidia GPU, uh, capabilities. So everything that's available on NGC cloud is actually available as runtime services and inference engines on top of our Kubernetes infrastructure. And we deliver with that, uh, you know, a, a rag ready database so that you get that outta box, uh, and delivered with the platform.

So, so that, I think it, it extends our VCF, it makes it much more capable, but it gives them all that application service that you need, and you can deploy it now onto a Kubernetes ready infrastructure just outta box. So, so kind of cool. I think on the AI side of things, it's very exciting.

We're seeing more and more of our service providers deliver, uh, AI services, uh, as a service. Um, because what you're seeing is more concern around, uh, data provenance and data locality, and it's leading to, uh, you know, kind of concerns around, um, uh, sovereignty. Um, the more's more sovereign AI services are getting built by our service providers and by our big customers.

Particularly when you look outside of the US region and, and as close as Canada and Mexico, but also of course, Europe and, and Asia Pacific Asia, I was just about in Europe. And you're seeing more and more ai, uh, services delivered in region, uh, as sovereign ai. Yeah.

I, I think that's gonna be a huge piece of it. Yeah. You know, Paul, in the, in the, in the tech world, we always focus on the latest, greatest shiniest and, and if only the world were full of green fields.

But you know what, not every app is gonna be an AI app. Not every app is gonna be a microservice architected app. Not every app is modernized, optimized for Kubernetes, and it's not gonna be that way probably for as long as you and I are working, there's always, there's gonna be legacy equipment and that the same, the same way.

We don't wanna have 12 different platforms that I gotta manage 12 individually. I don't wanna have one platform for the cool apps and one platform for the old apps. I need a platform.

Talk to me, why VMware's my best choice For that? Yeah. Well think, well, I'll go back and think about what you do off in, uh, uh, AWS cloud.

You have, uh, uh, VMs, that's what you get from, uh, uh, as you deploy, uh, your images and your, your your VMA images. Those are, those are VMs. And you've got a container service, right, which is the EKS service.

That's exactly what we do. You want a platform that does both. And actually in the VCF platform, you get VMs, you get container services, the container service, you can deploy basically what we call a cluster.

It's a set of set of nodes. That cluster actually has, like I said, all the Kubernetes services that you need. So you can have a cluster, which is actually your Kubernetes runtime cluster, just like you get up on Google, GKE, and you can have a set of VMs as a set of services that you run.

Uh, and those two can even coexist. You can actually run VM services. We've got something called VM service to allow you to do a full, uh, UB cuddle kind of, uh, desired state way of deploying VMs and services through Kubernetes as well.

So yes, I think developers want one platform because they have applications that are VMs and container based, uh, and you need both. Uh, but also on the infrastructure side, they want to be able to manage that as one single. The, the idea of having some separate silos, kind of container based application, separate from the way you manage the rest of the infrastructure, separate from your, your ai, bare metal service that you may have it, it's kind of a silly way to go.

Um, uh, I shouldn't say that to the IT people out there, but I think they know it. Um, uh, I think, I think unification of it will, will happen. It's inevitable because everything comes down to cost.

And the most cost efficient way to do that is to unify and simplify. Absolutely. One other area I want to bring up, I know we're running low on time.

We, we were talking off camera about RSA. You mentioned you hadn't been there in a while. It was huge this year.

Mm-hmm. It was huge because security and compliance is really no kidding around become top of mind for everyone can't afford not for it to be, for a prolonged time. VMware, one of their selling points has been, we're going to give you the most secure hypervisor, you know, virtualized environment.

We have that security and compliance built in equally is true, I think, for cloud native. Mm-hmm. Right.

Kubernetes and stuff. If you can enunciate what some of those security and compliance advantages are from hosting your cobe K eights on, on a, on a, uh, a VMware based platform. Yeah.

Well, we think it, it just the infrastructure side of things. We take care of everything from, you know, secure runtime on the CPU, you know, whether our integration in with a MD and Intel in terms of adev. So you can run kind of full confidential compute and runtime, uh, encryption of over the network encryption of the storage.

So the entire platform is actually built as a secure platform. Uh, and you've got the, you know, that that's the provenance of the way we build our platform too. We're actually very careful of our supply chain.

So we make sure that this is actually the most secure, trusted platform that you have. So, so the platform you can believe and trust in it. We also go a lot further though, at operational runtime, you actually want to kind of continuously monitor and check for anything that may be go wrong.

So because we virtualize the network, we can do some pretty cool things around, of course, firewalling. So you can start segmenting your network, get isolation. That's a very good way to actually get, uh, but to, to get protection in your network.

But we can also do, uh, kind of smart and intelligent discovery of attack vectors that are happening within the network. Uh, and that's the I-D-S-I-P-S capability. And so we're continuously monitoring for what could go wrong, uh, in the network, in your operations.

And then if things do go wrong, we've actually got the ability, I don't know if you're aware of this, but because we can do these kinda rapid snapshots, we can actually go back in time and unwind anything if you had a ransomware attack and actually bring back a clean version of things. So we are, I think, the most secure platform that you can run your application on. It's, it's a critical part of, it's what, as you said, what we've always been known for, and it's what something that, uh, we deem as, uh, you know, you always have your top three things that you must do, right?

It is always your top three as product management. Uh, security is always one of those three. It's, it has not changed for VMware, it was the same in version seven when I was doing planning for that.

It was the same in version nine when we were doing planning for that. Um, uh, we'll, we watch this one really carefully. We do fun stuff like going off and doing, you know, red teams out there to go and attack our product and see what's wrong with it.

Uh, uh, we're gonna keep it as secure as possible. Thank you. Hey, Paul, one last thing.

How can people stay on top of what's going on? What's the latest, greatest around VCF? Yeah, around VMware.

Yeah. Well, probably the best is, uh, two things I would say. Watch our blogs.

Um, VMware is probably the most, uh, blog proficient company you'll get out there. That's how we communicate lots of stuff. Like, yeah, just, just get to know us.

Uh, but otherwise, I would say VM mugs, every place in the country, every place in the world has v UG chapters. Uh, we've more than 120,000 members that are actually part of VM mugs. It's a, it's a wonderful community.

It's a place for you to learn. It's completely, you know, free all of these community events. You can actually just go learn in the local region.

You'll learn about the latest updates of the product. You can see what they have, but you'll also learn from each other, right? These are run by, these are user communities that are actually sharing experiences back and forth.

So, so get involved, get to know the V muggs. All the training is available, you know, for, for customers, uh, for free. So that's a great way to do it.

So you can, you can become a proficient VCF expert, uh, as, as part of that. And, and then the last thing that I would just say is, and, and also you get licenses by the way, uh, uh, if you, if you're trained in that way, uh, but get involved in that way. And then come join us at explorer.

You know, we've got explorer coming up in August for those in the us. Um, we're gonna actually run explorers across the world, you know, watch it in London, in Frankfurt, in, uh, Paris, in Tokyo, in, uh, in Australia, uh, in, in India. Um, so, you know, watch that second half of this year, you're gonna see us on the road a lot.

Um, but I'd say most importantly, look for your V mug. Uh, find your, find your chapter and get involved. I love it.

Paul Turner, thanks so much for coming on here today. I know we took a lot of your time. I appreciate it.

I know time's not easy. We're all scrambling. Continued success with VCF do keep us appraised though of what's happening.

Our audience is interested in this. Thank you so much. I'm Alan Schimmel for Techron.