Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Unleashing the power of KubeVirt - Running Containers and VMs on Kubernetes
2k Plays1 year ago
In this episode of Kubernetes Bytes, Ryan and Bhavin sit down with Sachin Mullick and Peter Lauterbach - the Product Management team at Red Hat focused on Red Hat OpenShift Virtualization and the open-source KubeVirt project and talk about how users can run containers and virtual machines side-by-side on the same Kubernetes cluster. They discuss the benefits of having a unified control plane for all your applications and the different features that enable users to run their applications in production. They also talk about some customers that have implemented this technology in production. Listen to learn more about how you can get started with KubeVirt and run your VMs alongside your Kubernetes pods on your Kubernetes or OpenShift clusters.
- 03:27 - News Segment
- 13:54 - KubeVirt Interview
- 01:06:12 - Takeaways
The Motley Fool:
Save $110 off the full list price of Stock Advisor for your first year, go to http://www.fool.com/kubernetesbytes and start your investing journey today!
*$110 discount off of $199 per year list price. Membership will renew annually at the then current list pric
Show Notes:
1. Kube by Example - https://kubebyexample.com/
2. Ask An OpenShift Admin - https://youtube.com/playlist?list=PLaR6Rq6Z4IqdsG6b09q4QIv_Yq5fNL7zh
3. https://kubevirt.io/
4. https://www.redhat.com/en/technologies/cloud-computing/openshift/virtualization
Cloud-Native News:
1. New Security Startup - Stacklok - https://techcrunch.com/2023/05/17/kubernetes-and-sigstore-founders-raise-17-5m-to-launch-software-supply-chain-startup-stacklok/
2. Traefik Lab announces Traefik Hub - Also raised $11M https://techcrunch.com/2023/05/17/traefik-labs-launches-traefik-hub-a-kubernetes-native-api-management-service/
3. KSOC releases the KBOM standard - https://tech.einnews.com/pr_news/629861155/ksoc-releases-the-first-kubernetes-bill-of-materials-kbom-standard
4. Upbound announces managed Crossplane service - https://www.infoq.com/news/2023/05/upbound-managed-control-plane/
5. Kubernetes 1.27 StatefulSet auto deletion for PVCs to beta https://kubernetes.io/blog/2023/05/04/kubernetes-1-27-statefulset-pvc-auto-deletion-beta/
6. Cost reduction CAST AI company focuses on reducing compute costs running generative AI models on k8s https://siliconangle.com/2023/05/18/kubernetes-firm-cast-ai-adds-support-reducing-generative-ai-deployment-costs/
7. Vault secret store operator https://thenewstack.io/hashicorp-vault-operator-manages-kubernetes-secrets/
8. Managed Kafka or Run it yourself ? https://thenewstack.io/kafka-on-kubernetes-should-you-adopt-a-managed-solution/
9. Cool usecase - edge k8s - robots picking fruit - https://thenewstack.io/fruit-picking-robots-powered-by-kubernetes-on-the-edge/
10. Knative 1.10 release https://knative.dev/blog/releases/announcing-knative-v1-10-release/ (4-25 missed it)
Recommended
Transcript
Introduction to Kubernetes Bites
00:00:03
Speaker
You are listening to Kubernetes Bites, a podcast bringing you the latest from the world of cloud native data management. My name is Ryan Walner and I'm joined by Bob and Shaw coming to you from Boston, Massachusetts.
Cloud-native News & Data Management
00:00:14
Speaker
We'll be sharing our thoughts on recent cloud native news and talking to industry experts about their experiences and challenges managing the wealth of data in today's cloud native ecosystem.
00:00:29
Speaker
Good morning, good afternoon, and good evening wherever you are. We're coming to you from Boston, Massachusetts. Today is May 19th, 2023. Hope everyone is doing well and staying safe. Let's dive into it. Bobbin, before we get into today's topic and news and guests, what is going on? How you doing?
Mid-Year Reflections & Lawn Troubles
00:00:47
Speaker
I don't know, when you just said May 19th, I was like, shit, like we are halfway, almost halfway done with the year. And I need to go back and check my New Year's resolutions and what they were. We should have the day after. I know, but like half, half the year. Okay, yeah. Okay. No, I need. Okay. Pick up. Come on. Let's do it.
00:01:10
Speaker
I don't know the like, with the weather that it has been right. I did my first lawn mowing. I know people on the people who listen to this part know about my trials. That's great. That's great. Yeah, that's big. I hated it again. So I think I ignored it for a month. And now the weeds have grown and they're really difficult to pull out. So I'm leaning towards getting some professional help. So if you have any recommendations for
00:01:39
Speaker
Gods and Landscapers. Yeah. Send them over, you know, Ryan and all the listeners. I'm all yours. You can make it a community project and get the local kids over and know you're allowed. I'm very lazy compared to like how big the lawn is. So I know I need some help to at least get it in a good shape and then I can maybe maintain it. So day zero and I need help with day zero. I can do day two operations.
00:02:06
Speaker
Yeah, I saw I saw funny real the other day on Instagram that was like about people mulching and making their lawns look really nice. It was something like, you know, mulching what you do to hurt your back and make your lawn look good for a day. That's accurate.
00:02:23
Speaker
Yeah, I hate those reels, dude. Whenever you scroll by, people's lawns looking so awesome. They have these new papers and everything. I was like, damn it. That's so much work. Effort. It's tough. And I always felt like, unless you're staying at home constantly, I don't know. Which we do. Having a lot of people. I'm out of the house more often than not. Not worth it. Keep it up with the Joneses. It ain't worth it.
00:02:46
Speaker
Let it grow, you know, let it grow. Yeah, it is flying by.
Red Hat Summit & Startup Introduction
00:02:51
Speaker
I know we have Red Hat Summit next week, which, you know, we'll talk about a little bit with our guests, but you know, we'll both be there. That's exciting. And probably when this is out, it'll be at Red Hat Summit. So if you find me or Ryan, just say hi, I'm pretty sure we'll bring the same stickers. So we have some stickers for you, if you find us so for sure.
00:03:12
Speaker
Absolutely. And if you come find us and you're interested in coming on the show, we'll, we'll do that too. You know, we'll put something up. No worries there. Um, all right. Well, um, I think without that, with all that done, I think let's drive into the news a little bit. Yeah, let's do it. So I have a few, uh, like not announcements, but news to share first one of, uh, the list being, uh, a new startup, uh, with, with,
00:03:38
Speaker
a famous co-founder, I guess. So Craig McCluckie, he was the co-founder of Heptio. He has now started a new startup around the software supply chain security, along with the founder of the six-store project, Luke Hines. So both of them started a new startup called Stacklock. And again, they picked up Series A, $17.5 million, and they aim to add prominent data for developers to make the whole software supply chain secure.
00:04:03
Speaker
Again, kudos, kudos on the naming there. I'm naming Stack Lock, you know, it's security oriented, very good. Also, yes, famous in our neck of the woods. If you ask your grandmother, they will have no idea. But if they're listening, look up Stack Lock.
00:04:22
Speaker
No, I think they didn't do a seed since the amount of money raised was so high, they just went with a series A. So there was no seed, just series A. They just came out of stealth and they do have like four or five senior engineer positions open. So if you are looking for a change, yeah, there is a new startup in town. Open positions is tough in today's
K-BOM & Cost Reduction in AI
00:04:42
Speaker
market. So go get them.
00:04:43
Speaker
And then another one was Traffic Labs. They announced a couple of things. So they announced general availability of Traffic Hub, which is its cloud-native API management solution. The whole point here is you can plug and play different proxies, so a traffic's own proxy, but even Nginx, HAProxy, Ambassador, and others. So it gives you a unified management tool for your APIs, but then you can swap out different tools based on your requirements, I guess.
00:05:12
Speaker
But yeah, and then the last one being around the security space, K-SOP, which I thought was a standard because I think it's a Kubernetes Security Operations Center, but it's actually a startup. They released a new standard called Kubernetes Bill of Materials, or K-BOMs. I know we have spoken a lot about S-BOM. We did that with Christopher, the co-founder of Montu as well.
00:05:36
Speaker
But now we have a new acronym in town called KBOM. It basically provides a quick view of scope of your Kubernetes cluster, so it gives you workload count, cost and type of hosting service, vulnerabilities of both internal and hosted images, and so on and so forth. But a new standard that gives you a complete overview of what your Kubernetes cluster and applications look like. So yeah, that's news from me.
00:05:56
Speaker
Yeah, I know SBOMS we talked about before in the context of Kubernetes, but I think this one's designed very specifically for Kubernetes. I thought it was an official organization as well, not just startup, which it's probably on purpose. It's exactly what they want people to think. We are the standard. We are the company that should be building standards. Exactly.
00:06:17
Speaker
Cool. So I have a couple of things worth noting here. 127 is sort of like frozen or something. Is it out? I forget. Oh, it was out. Yeah, that's right. And the staple set auto deletion for PBC is in beta. So if you use staple sets, I think that's an interesting thing where you can kind of set up auto deletion rules, so to speak, about how often those things get cleaned up after the staple set doesn't use them and things like that.
00:06:47
Speaker
I kind of equate this to when you leave, and accidentally you leave your EC2 volumes around for a while and it costs you a lot. But they're costing you something. So of course, be careful with auto deletion. Of course, we don't want to delete your data.
00:07:06
Speaker
Um, some, uh, news around the AI space. I think, you know, there's a ton of news about AI here, but I focused in on this article because it's a company, um, called Cast AI, which focuses directly on reducing compute costs, running generated AI models on Kubernetes. So, um, I found this interesting because of.
00:07:28
Speaker
just the use case around chat GPT and generative AI models like Bard and others are so new and yet there's companies focused on reducing the cost to run these things. I feel like it was such a quick turnaround, just an interesting article about how running these things and keeping the compute available, if you ever interacted with chat GPT, you can see if you have a free one, so it's not always available. It's not easy to run these things or to make available to the masses.
00:07:57
Speaker
So I think some interesting problems as we start to adopt them more and more as a society. So a pretty cool article there. Yeah. And as we know, right, like open AI, like Microsoft maybe has a 49% share in open AI and that's because they gave them so many Azure credits to run the service. So yes, like running these things are expensive.
00:08:16
Speaker
Yeah, exactly. And this one specifically to like running on a Kubernetes, because, you know, that is a, I guess, a way in which the architectures are commonly being built. So and I think we covered the other funding round back in February. Yeah, go ahead.
Secret Management & Kafka Debate
00:08:32
Speaker
And, you know, I imagine there's going to be a plethora of AI companies announcing themselves, you know, from now until the next few years and probably beyond that, but
00:08:43
Speaker
All right, next one is if you've used HashiCorp Vault at all, you might be, you know, kicking yourself about managing the secrets that you use that for with Kubernetes. And they came out with an open source project called Vault Secret Store Operator. Operators are the future. We've talked about that. We've had entire episodes on operators early on, I think, in this podcast, which is probably worth another look. You know, I've seen where operators have come from.
00:09:09
Speaker
Yeah, for sure. This is one specifically that is, you know, given back by Hashicorp Vault and basically makes managing the Kubernetes secrets easier, which, you know, no one's going to, I think, disagree with making things easier. So go check it out.
00:09:24
Speaker
that out. The next article I have is around basically using managed Kafka or running it yourself. And I know I specifically picked this one. It's a new stack article, mostly because we've talked about running staple workloads on this podcast a lot.
00:09:40
Speaker
And then we've also talked about the trade-offs of whether you use managed Kubernetes and other managed services or databases as a service. I think this article does a really good job of the thought process behind why you'd want to do something like that and what you'd have to take on from a hiring standpoint.
00:10:00
Speaker
You need an SRE, you need a software engineer, you need an admin, those kind of things. And kind of weighing those pros and cons and the kind of company you are. So I think it's a good article. Go check it out. Definitely worth looking at.
00:10:16
Speaker
And then I wanted to put a couple or one release in here, which is Canadian 110. I'm not going to go into what is in 110, but we've talked about Canadian on here past. So 110 is released. And that was actually a few weeks ago. We just missed it. And that was a cold start problem.
00:10:38
Speaker
And the last one I have on here is I might start doing this, like putting cool use cases in our news article. Anyway, cool use case of the week, I'll call it, is an edge Kubernetes deployment of robots picking fruit.
Robotic Fruit Picking & K-native Update
00:10:52
Speaker
It's just a very, very interesting use case around basically solving a problem of fruit not being picked in time and basically controlling waste.
00:11:00
Speaker
in farming and fruit and that kind of thing. But it's Kubernetes at the edge. So it's a very, I think when I read a story about infrastructure from a technology side about a real kind of interesting use case about solving a real problem, it always kind of connects things better for me. So maybe I'll call this a cool use case of the week segment or something like that.
00:11:20
Speaker
But go check that out, really cool article, and that's all I have for the news. So with that, we have some awesome guests
Qvert Overview & Use Cases
00:11:29
Speaker
today. We're going to be talking about Qvert. I know, Bob, and you brought these guests to us and I'm excited for this. Qvert is a very popular sort of topic today about running virtual machines.
00:11:41
Speaker
in Kubernetes or next to Kubernetes or next to your container. So we're going to dive all about that into Qvert with Sachin and Peter. And without further ado, let's get them on the show. We'll be right back after this short break. Look, I don't know about you, but inflation has been a doozy this past year, year and a half. And just in April, motor vehicle repairs way up, frozen vegetables, if you're into that kind of thing, and pet food. I have a Golden Retriever at home.
00:12:10
Speaker
I can't keep doing this, or at least I have to be better with my finances. And look, we get it. The market is complicated and confusing. And to many of us, it simply doesn't make sense. In fact, where do we even start? Take all of the guesswork out of it.
00:12:25
Speaker
with The Motley Fool's Stock Advisor. The Motley Fool has been around for over 25 years and has been spot on in recommending some of the world's most important companies before they hit big time. I'm talking about Amazon, Tesla, Netflix, and Starbucks all before they exploded in value. And they're easy to use and super informative service Stock Advisor. You can join the ranks before they potentially find the big thing.
00:12:55
Speaker
After all, their average stock recommendation is up over 400%, and that's of April 10th, 2023. And no need to be intimidated by the financial jargon or market complexities, as the name suggests, these guys don't make themselves too serious.
00:13:12
Speaker
Now, finances, that's a different story. Their friendly and relaxed approach has helped over 700,000 people move closer to their financial independence, all while beating the market and having fun. New members can access the stock advisor for only $89 for the first year, a full $110 off the full list price.
00:13:32
Speaker
Don't sit on the sidelines and think about what could have happened. Visit fool.com slash podcast and use code Kubernetes Bites or visit the link in the show notes. This is $110 discount off the $199 per year list price. Members will renew annually at the current list price and until then Kubernetes Bites send you.
00:13:52
Speaker
And we're back. All right. Welcome to Kubernetes Bites. Sachin and Peter, it's great to have you on here. Why don't you introduce yourselves for our audience here? Sachin, why don't you go first? My name is Sachin Malik. I work at Red Hat. I manage the product management team. That's about all things virtualization. Our primary product is based off of Kube word called OpenShift Virtualization. That's what we'll be talking about today.
00:14:22
Speaker
Right. How about you, Peter? Sure. I'm Peter Lauterweg. I'm one of the product managers here on the OpenShift team, OnSachin theme. Focus on a couple of different things here at Red Hat. Traditional virtualization or Red Hat virtualization, Rev, OpenShift on Rev, which is a deployment option that some customers use, and then OpenShift virtualization or Upstream Kubernet.
00:14:46
Speaker
In a former life, I actually did a stint at a couple of different storage companies and different startups. So I've been in kind of the virtualization space for probably actually at the time that the EMC bought VMware.
00:15:01
Speaker
Oh, there's a lot a lot of crossover here too. We were just talking with Sasha about I was there in 2012 and I left before the Dell and EMC and everything. Everybody asked me did it did I know and I because I left like the week before it announced, but I did not.
00:15:20
Speaker
But it was an interesting thing. One of my buddies who I respect, and I believe he's still there, said, hey, we just bought this VMware thing, and it's really cool. And he explained software virtual machines to me, and I'm like, I don't get it. He's like, just download, what was it, the workstation at the time. And I did that, and holy cow, within an hour I had Windows, Linux, and FreeBSD running on my workstation. It was amazing.
00:15:50
Speaker
That's awesome. Yeah. And Ryan to talk about, I don't know, connections, right? I've reported to two of the people on this call right now. So I used to work for Sachin over at Lenovo. And then I know Ryan for some period at Portworx, I did report to you. So I have two managers. Peter, maybe you're next. You never know.
00:16:12
Speaker
No, OK, so we're here to talk about Qbird, right? So why not start with that 10,000 foot view? Like what is Qbird? What it's used for? What are some of the use cases? Let's start there and then we can go into a bit more detail. OK, I can I can start. So Qbird is actually the upstream project for our product or the OpenShift virtualization. As you know, everything that Red Hat does has an upstream community and project associated with it. Yeah.
00:16:40
Speaker
And what what Qvert is is actually we've taught Kubernetes how to run KVM virtual machines. It's actually pretty simple. Right. So, you know, Kubernetes is great with containers and does all sorts of good things. And like, I don't think I need to explain anybody about that, but
00:16:57
Speaker
VMs and virtualization, there's still a lot of gravity for both data and a lot of business logic that's running inside of virtual machines. And you can either have a choice of, hey, let's just leave this behind and shed it and go directly to everything in containers, which, by the way, is completely unrealistic, or you bring virtual machines along with you.
00:17:19
Speaker
Right. And the idea of there's a couple of different use cases, which is, hey, I've got an existing three tier application, you know, database middleware web front end, and I actually want to modernize it. Right. So I want to be able to, you know, hit it with a hammer, put maybe a mobile front end on it, take my middleware and break it up into more microservice stuff. That's one of them.
00:17:43
Speaker
The other one is I have an application delivery platform that's based on virtualization today, and then I want to be able to do containers as well. You can run your container platform on your virtual platform, and a lot of people do that, but now you really have to manage two platforms. Can you connect applications across the two? Sure, but you now own that and that's your job.
00:18:12
Speaker
Now you have application delivery, say a composite application that has VMs and containers all running on the same platform. We'll probably talk a little bit about the details of how that happens. But virtual machines and Kubernetes use all of the Kubernetes abstractions, services, routes, things like that. It's actually very easy to connect virtual machines and containers together.
00:18:40
Speaker
And then the last one is, you know, there are some applications that run inside of EMs that
00:18:46
Speaker
they're never going to be containerized, right? In fact, it might be a windows.net that the guy that wrote it is retired and nobody wants to touch it and there's no need to. And that's fine. You could just migrate and replatform that application because it's still running on a core piece of your business and run it on Kubernetes alongside. So now you have one platform that you're going to manage, but you can actually run a mix of workloads inside of it.
00:19:14
Speaker
I think this falls under the teaching a new dog old tricks category.
00:19:20
Speaker
Absolutely. Well, one of the things, if you think about it, Kubernetes is awesome with containers, but it doesn't do certain, it didn't do up until recently, a bunch of stuff like it didn't know how to manage metal, right? You always ran Kubernetes in a cloud somewhere. So we actually took, you know, Red Hat as a company, we took a lot of the innovation we did for say, ironic, which is metal management. And we taught Kubernetes how to manage bare metal, right?
00:19:48
Speaker
Qvert is the same thing. We've taught KVM is a very mature technology that's been around for what, 12 years? It's been in the Linux kernel for quite some time. We're just going to teach Kubernetes how to speak to KVM, you know, KVM. Yeah. And, you know, the next the next one is OVN, you know, OVN, right, software defined networking and Linux runs on multiple platforms. But guess what? It also runs on OpenShift.
00:20:17
Speaker
Nice. Okay. Those are some clearly defined use cases. Thank you for that. One thing that I wanted to ask since we have the experts on, a lot of people in the community think about Qvert as a replacement to VMware. They're like, okay, if I started opting Qvert, I can maybe go away from VMware vSphere. Is that the case? I know vendors like Nutanix and Microsoft and even Red Hat have tried
00:20:42
Speaker
moving people away from vSphere to a different virtualization stack. Do you see that happening? Is that even possible? What are some of the thoughts there? Yes, sure. Actually, it's a hot topic of the day with a lot of things changing in the industry.
00:20:57
Speaker
including Red Hat actually moving away from some of its traditional virtualization products. There's a reason why we decided to do that because we feel the value to a customer is not in the hypervisor layer, right? So that's the value that started in the 90s like Peter was describing with the workstation where you become independent of your server or your desktop machine and you can earn multiple of it. So
00:21:25
Speaker
If the reason Kubernetes exists is now people started seeing the value is actually coming by doing things in a more cloud native way. And it's coming from automation, declarative way of managing using the Kubernetes principles.
00:21:42
Speaker
And what we are trying to do is, as Peter said, the new applications are most likely being written in Kubernetes. Like, my son comes out of college, that's how he's gonna do all the stuff coming up. This around if you'll believe like the
00:21:59
Speaker
big companies that do this research, like 77% of the IT workload is still in VMs. It will go down to like 65% by 2025. So two-thirds of your applications are going to be virtual machines. So what we are trying to do is
00:22:17
Speaker
help people have a journey where they can take this stuff that's not going away in one year, two year, probably a 10 to 15 year journey. Get start getting familiar with the new way of doing things because the VMs will be what they're familiar with, right? You have to do a start, stop, restart, snapshot, all the same things. The VM that somebody does still has to happen.
00:22:42
Speaker
But then you can start adding principles like GitOps, Tecton, Pipelines. And then you have a way of managing things that you never had before. So that's the journey we want people to be on. We don't think people can take their existing platform and replace it in one day.
00:23:02
Speaker
It's going to take couple years, five years, but do they want to stick to what they have today? That's going to be probably very difficult. If they start looking three years ahead, they may not be making the right decisions today. So that's why we created this technology.
00:23:19
Speaker
Yeah, and we've had a few other guests where we've talked about the topic of sort of, what does the future look like? And I think in every one of those instances, it always includes virtual machines, right? But then it also takes on sort of new technologies like web assembly, right? And kind of enables you to run sort of the right tool and use the right tool for the right job. I think we're seeing that trend of, you know, development teams and those kinds of things.
00:23:49
Speaker
want to be able to use the tools for the job. And to Peter's point before, right, you know, you might have an application where like, don't break it, it's not broken, right? Just leave it. So I think this rings true. And, you know, maybe this is a good lead in to, you know, both of
00:24:04
Speaker
you know your previous answers both Sasha and Peter which is, you know, why are organizations, you know, thinking about making this move, is it, is it really to kind of stay up with the way the industry is going with Kubernetes like you know do they want to
00:24:20
Speaker
have a platform in this new platform engineering term that we're thinking about that can support virtual teams and VMs or what's really the value of moving away from a tool like VMware to something like OpenShift with VMs that you're maybe seeing the most.
00:24:38
Speaker
Yes, so I think we talked a lot about it, but if I have to sum it up, so the number one thing, like if you think about anything is things start out as products and they become features over time. Yeah, that's what we feel like now we have a unified application platform and you can take examples from your day to day world, right? So you don't need to manage silos. You have just one thing to manage, so that attracts people a lot.
00:25:07
Speaker
The other one is there aren't like clear mandates to companies about modernizing their infrastructure, going into a cloud native world, going into hybrid. And as they look at that, again, as I said, like if they only look at greenfield applications.
00:25:26
Speaker
that doesn't give them much value, right? So that's why they're going back and looking at what exists today. And then I do feel like some of the things that happen in the industry where clearly there wasn't enough value for virtualization players just to be pure virtualization players and they're starting to get bought out.
00:25:48
Speaker
So the companies that rely on them are getting a bit edgy. Like, am I really doing the right thing? Is this an opportunity? Like, switching has a cost, right? So if I just even go from Android to Apple, it's going to make me learn a bunch of new stuff. So why would I do that is
00:26:09
Speaker
For whatever reason it is, when I go to a new platform, I want to make sure I'm not just doing it just to save a few dollars. I am with something that's going to keep me happy for the next 20 years. So that's what we are seeing.
00:26:25
Speaker
That makes sense, right? The value has to be worth it. The value of the new solution has to be worth it. And I agree, right? There is benefits to having a unified platform where you can manage VMs and containers and follow these new principles that you guys discussed, right?
00:26:42
Speaker
If I want to use GitOps for my new applications, why not try to implement some of that into my virtualization layer as well and see if it works for my VM workloads. But okay, so now let's go into those details that we promised. If I could expand on that a little bit. Yeah, please. It actually is about how you use it, not about the technology, right? Anybody that thinks, oh, I'm going to run VMs in containers and my life will be better.
00:27:09
Speaker
Yeah, exactly. But let me give you a little history lesson, right, which is agile software development for developers has been going on for what, close to a decade now, right? It started with Naven and Gradle and, you know, Java-ish kind of stuff. But the idea of a fast iterative loop that's been around for a while now, right?
00:27:29
Speaker
Kubernetes just makes it easier. So all the software developer guys have had these benefits for over a decade now. The infrastructure people are stuck with pretty much what they built by hand in their vSphere environments, right? Which, yes, there are other virtualization platforms and technology, but let's face it, right? That's the standard.
00:27:53
Speaker
But now, okay, I want to use a declarative way to roll out, you know, using something like zero touch provisioning, which is the ability that it's essentially GitOps to roll out hardware, which is right now it's all telco and edge stuff. But you know what, you can use that same technology to lay out a rack full of data center. And
00:28:15
Speaker
You know, everything's declarative. It doesn't drift. And if you want to change something, you check code in, it gets reviewed and automatically tested and rolled out. Yeah. And I think that's a different way of a different way of operating. Yeah.
00:28:32
Speaker
I absolutely agree. And I even would say we've seen a trend with a lot of newer startups that are kind of taking both the infrastructure piece and the software piece now and kind of cramming them together in a single abstraction where they basically let you point at what you want to deploy and they'll go ahead and take care of everything from
00:28:52
Speaker
you know, zero-touch provisioning to deploying your app as well instead of a single sort of control plane, which, you know, really just drives, I think, at, again, the goal of faster loops, right? And ultimately, to your point, it really comes back to still how you organized yourself as a team, right? And culture and being able to support that kind of rollout. Right. And let me use the example of releases, right? It's not just faster loops, it's smaller increments, right?
00:29:22
Speaker
A traditional virtualization platform, your life cycle, major releases are every 18 months to two years and a couple of months in between patches, right? Something like Kubernetes drops, actually used to drop four times a year and people are like, okay, slow it down.
00:29:39
Speaker
So Kubernetes drops three times a year. And then on the OpenShift side, we do Z streams weekly, right? So every week, there's, you know, hey, there's a dozen bug fixes in this, and you can take them or not. But yeah, just rolling forward on an incremental basis is literally implicit to the platform. Yeah.
00:30:01
Speaker
Makes sense. Okay. So now let me get
Technical Deep Dive into Qvert
00:30:04
Speaker
into the details. Like I know we promised our listeners that, right? So how does Q but actually work? Like if I, we all know that it runs on Kubernetes, but does it have its own custom resources or CRDs available? Does it run as a pod or there's a new VM object like Peter can, or such. And like, can you talk about like how it actually works?
00:30:23
Speaker
I can. And this is the interesting part because people are like, well, KVM, VMs and pods and all that. Do I have to create a container? Put like, no. So virtual machines and Kubernetes are good citizens of the Kubernetes platform. So they're first class objects. Yes, we do use CRDs and we use an operator, but you operate on virtual machines. You say, create a VM from a template.
00:30:52
Speaker
put these kind of disks in it, you know, set it this kind of memory. Everything is familiar to you. And then the Kubernetes stuff happens underneath.
00:31:04
Speaker
Let's use two basic examples. How do you get your storage? In the old world, let me compare and contrast, you would get a massive lawn from your storage team. You put in a ticket a couple of weeks later, you'd get this lawn and you put it in and then you'd put the data store on it.
00:31:24
Speaker
Now, you as the admin have to pay attention. How many VMs do I put in that data store? Am I running out of space? Is it performant enough, right? That's barbaric, right? Like, I don't want to do that. I just want to manage. I was going to say, people are cringing at the fact that you said two weeks later, you get your your lawn. Actually, I won't name the company.
00:31:47
Speaker
We've had a customer say to us, look, I need you to be able to zero this out when I want to, when I unallocate it. And we're like, why? Just destroy it. And he's like, no, because once a storage team gives me the lung, I never give it back to them. I just have to wipe it out and then I reuse it. It becomes your pet. Yeah, like, okay, so you want me to put in a product feature because of your organizational deficiency?
00:32:16
Speaker
of the good one there. So now let me tell you how it does work in Kubernetes. Kubernetes actually has a very strong storage abstraction called CSI, the container storage interface. And what's very cool about that is it's a set of APIs and definitions that say, okay, here's how you provision storage, you get persistent volumes, and you get them through persistent volume claims, right?
00:32:42
Speaker
So in our world, it's a one-to-one relationship. So every virtual machine disk is a PV that's claimed through a PVC, and you don't have to worry about it, right? You get that it's self, Kubernetes is self-service, right? So I go and I instantiate this virtual machine. It will go request it from whatever storage classes. And this is the other cool thing. Kubernetes has this idea
00:33:09
Speaker
which, you know, if you compared it to something traditional, it would be like storage-based policy management, which is, it's declarative. Tell me, is this fast storage? Is this bronze, silver, gold? How often is it backed up? Right. Well, Kubernetes has that. You say, let's use an example, say, Adelare, right, where you say, hey, and by the way, there's like many to choose from.
00:33:35
Speaker
but let's just pick one like PowerFlex, which is the new software to find stuff, right? You can go create a couple of different storage pools within that and expose them as storage classes in Kubernetes. And then when you create your VM, we'll either use the default one or whatever storage class you tell us.
00:33:54
Speaker
Now a couple of things happen there. One is storage is no longer your problem as a VI administrator because how much capacity is in that storage pool? How is it replicated? How redundant is it? How is it backed up? That's not your problem anymore. There's a professional that's managing the storage part of the Kubernetes cluster. You as a VI admin don't have to worry about any of that anymore.
00:34:22
Speaker
So networking is very, let me finish on the storage side. So there's really two classes, right? One is your traditional vendors, right? NetApp, Hitachi, Dell, Hewlett Packard, and gosh, I'm probably leaving somebody up.
00:34:37
Speaker
All of the traditional array vendors have CSI operators and drivers that connect directly to their storage. So iSCSI fiber channel, NFS. And that's one class, and they work fine with virtual machines. So Qvert,
00:34:55
Speaker
does that. And then there's another class which I would consider cloud native, right? And that could be something that either runs inside the Kubernetes cluster, like say Portworx, or LinStore. There's a couple of storage OSs, another one. And then OpenShift Data Foundation, or you can actually have an external cluster like an external Ceph cluster that you can expose storage into.
00:35:24
Speaker
Once you make those connections into the Kubernetes cluster with a storage class, you can expose request data, request volumes from it directly. So networking is very similar, right?
00:35:40
Speaker
There's actually multiple choices for networking. On the OpenShift side, you can use the OpenShift SDN, which was our default up until a while ago. Now we've released OVN Kubernetes, which again, that was the
00:35:59
Speaker
essentially the networking stack from OpenStack, but it's very capable. It's now the default for OpenShift, and you can you can deliver your and plumb your virtual machines into that. The very cool part is Linux networking is very capable
00:36:17
Speaker
And in the past, you know, a lot of IF config and a lot of, it was just messy, right? And at some point somebody said, Hey, we should make this declarative. And this was even before Kubernetes. And they said, we're going to create network manager, which is exactly that. It's a declarative way to say, I take these physical things that you find and carve them up this way, VLANs, bridging or bonding. It's all declarative.
00:36:42
Speaker
And it turns out there's actually an operator for that, right? So you can actually teach Kubernetes how to use network manager to essentially configure your entire network in the cluster, right? And if you don't like either one of those, you can pick one of the CNI vendors, right? So let's see, there's Cisco ACI has a CNI interface, Tigericalico, Isovalent, Thilium,
00:37:12
Speaker
And this is the cool part is like, now that you buy this thing, they all have interactions into Kubernetes, you don't have to do it. You just go into the catalog, download the operator and tell it how you want it to be configured and boom, your Kubernetes can speak, you know, isovalent.
00:37:34
Speaker
I know that's a great part about having standards, right? Like you have that, you spoke about the CSI standard, as long as every vendor complies with that, it works out of the box with not just pods, but also a queue, but VMs. Same thing with CNI. As long as you are complying with the CNI standard, you can bring in your Calicos or Cilium meshes. And then if you wanted to do, I didn't know Cisco ACI had a CNI plugin. So that's something new I learned today.
00:37:57
Speaker
They absolutely do. And now the interesting part is the virtual machines are part of the Kubernetes end-to-end testing, and some of the testing is more complete than others, right? So in the CNI testing, we actually spin up two VMs and we connect them together. The storage one, we need a little bit of work there. I do want to touch one more thing on the VM part. Sure, yeah.
00:38:23
Speaker
So the interesting part is KVM is super capable, right? You can create a VM with hundreds of vCPUs and terabytes of memory and stick something like SAP HANA inside of it. You can do that on multiple Red Hat platforms. Those capabilities exist on OpenShift, right? So from a performance point of view, for a similarly configured hardware, you should expect performance parity for running something like on RHEL or Rev.
00:38:53
Speaker
Right?
00:38:54
Speaker
And then if you need to do anything like, I've got a fat NUMA server, right? And I've got a database that I want to map, you know, map to memory, map it across a couple of different NUMA zones. You absolutely can do that as well. GPU enabled stuff, you got, you know, you got big fat GPUs in your servers that you want to connect to your virtual machines. You can either do that directly through PCI pass through, or there's actually a VGPU operator that can actually slice up a big fat GPU.
00:39:24
Speaker
into different virtual machines. So a lot of the capabilities, the Q-Mute KVM capabilities that exist on other platforms also exist on OpenShift.
00:39:37
Speaker
Got it. Yeah, that's, I mean, that's super powerful. And I want, you know, I want to take a stab at sort of reconnecting all the dots that you just laid out for us a little bit. I mean, you correct me if I'm wrong, but Qvert as a technology, that gets basically installed to the base nodes, right, where applications are going to run. And then you have a networking and storage component that networking, as you said,
00:40:01
Speaker
allows the VM to communicate with other containers, with, you know, get hooked up to services and OpenShift or Kubernetes and that kind of thing. And then the storage component, I'm guessing, has both sort of the concept of a root disk, which is like your operating system and then sort of potentially also using the same CSI components for a data disk. So if you have an application in your VM that needed space to write, then you have those components.
00:40:29
Speaker
So there's sort of an infrastructure piece that is installed and then you have sort of the networking storage components that are kind of also tied together with OpenShift or Kubernetes. Is that sort of summing up? That's exactly right. Let me make a minor technology changer.
00:40:46
Speaker
So Qvert requires bare metal servers. So what actually happens is just as rel is a KVM hypervisor and rev is a KVM hypervisor and OpenStack is a hypervisor, all types zero. The Kubernetes node running on bare metal is the hypervisor. So there's nothing to install, right? Linux already knows how to run VMs, like that comes that way out of the box.
00:41:15
Speaker
When you install the Qvert operator, what you're installing is the CRDs and the APIs to actually manage and create VMs on that platform. But the actual technology to run KVM is literally baked into the Kubernetes node.
00:41:31
Speaker
Right. Okay. So talking about, like, I'm glad that you brought up terms like new ones, new ones, because I'm pretty sure like VMware admins that have gone to VMworlds for years, definitely will feel more comfortable now that you have listed those technologies.
00:41:49
Speaker
especially the VGPU operator and splicing up your GPUs, that is awesome. One small follow-up question was, does this support all guest OS's? Can I do Windows Server? I know obviously we can do Linux space, but my question was around Windows Server. Can we bring on those workloads as well?
00:42:07
Speaker
Absolutely. So KVM can actually run Windows, Linux, you know, rel, Ubuntu, Sentos, you know, whatever flavor you want. It'll run there. The only, I wouldn't say trick, but the only thing you need to remember is Cumul actually supports two different machine types. There's the older i440 fx, which I don't know why I named that, but it's
00:42:33
Speaker
but we'll put it in the show notes, don't we? You've actually only managed, works with Q35, which is a more modern virtual machine hardware type. Sure. And that's what you want to be using anyway. So as long as you can deploy your operating, install your operating system on Q35 machine type, which the answer is almost always yes. You can run Windows, you know, now here's the interesting part.
00:43:04
Speaker
windows at Red Hat.
00:43:08
Speaker
you know, we validate the platform as part of OpenShift, right? So we actually go through the Microsoft has a validation program. It's called the Windows Server Virtual Validation Program, SPVP. And we do that for every Red Hat platform. So, RHEL, Rev, OpenStack, and now OpenShift. So we run through that entire test suite in Windows VMs on OpenShift. It's actually a release criteria for us.
00:43:35
Speaker
passes and you can actually go to the Microsoft site and say, look, OpenShift is validated. And what that means from a commercial point of view is, you know, as long as you're running Windows Server 2012 R2 and later,
00:43:50
Speaker
and you have a problem with it, you know, and that's like, you know, SharePoint, SQL Server, you know, Windows, desktops, 10, 11. And you have a problem, you can either call Microsoft or you can call Red Hat, either one, and you're fully supported.
00:44:08
Speaker
Got it. Yeah, that's, that is very powerful. And I feel like, you know, the more that we get into this, the more it starts to feel like, you know, that you could easily see this as a reason, right, if you were on the edge before of should I be running VMs in my Kubernetes infrastructure or OpenShift?
00:44:27
Speaker
infrastructure, it sounds more and more like something I'm used to in sort of like, if I was a VMware, sort of used to using those. And I think on that note, since we haven't kind of dove into sort of the beyond day two, or sort of, you know, the basic use case is, you know, if I'm used to using sort of vSphere and things like vMotion, or vSphere HA, you know, how do, you know, the OpenShift and Kubernetes sort of
00:44:55
Speaker
covert integrations line up with those types of features that might be seen as sort of, you know, advanced, you know, topics. Right. Well, the emotion or live migration is, is not advanced. It's actually literally
00:45:12
Speaker
It's table stakes for a virtual platform. If you can't do that, you can't call yourself a virtualization platform. But as most people say, hey, but if you're running a pod on a host and you want to run it somewhere else, you don't move the pod, you just yell it and start it somewhere else.
00:45:31
Speaker
So we actually do that for you. I remember when I said all of the Kubernetes-ness is handled for you. So you say, hey, I want to migrate this VM from this node to that node. We'll actually start, and it's the same technology that you use on other platforms, right? We will start a pod on another target node and we will start streaming the running VM and the memory over there.
00:45:56
Speaker
you'll need shared storage. So you need to be able to connect the PVs on the different hosts and we'll stream it over there while the VM continues to run. So then this is how live migration works on every Linux platform. And when you're ready to go and all the memory and data is over there, it's converged, then we'll actually stun the VM,
00:46:19
Speaker
start running it on the target system and then delete and disconnect the pod over there. And obviously the storage has moved over, right?
00:46:29
Speaker
the cumucavian process is the piece that handles the, the actual mutex of who's writing to the disk. Yep. So the cool part is, is I've just described something that's actually incredibly technical, but you can migrate multiple VMs at a time, you can monitor their progress, you can cancel all migration at any time. But again, in Kubernetes style,
00:46:53
Speaker
you as an admin don't need to worry about that. So say I want to upgrade my Kubernetes cluster, right? And over-the-air updates will automatically roll out through your cluster. The Kubernetes upgrade operator will go, hey, I need to cordon off and drain this node, right?
00:47:10
Speaker
and every VM will get a notice like you need to be running somewhere else. And as long as the live migrations, the run strategy is set to live migrate, it'll exactly start doing that in batches if you wish. And then when all the VMs are drained off of that, it'll upgrade that node, reboot it, and it'll come back up and rejoin the cluster. And you'll do that through every node, every worker node in your Kubernetes cluster.
00:47:39
Speaker
That's awesome. I think it's a good citizen of the Kubernetes platform. To prep for this episode, right, I was going through and asked an OpenShift admin YouTube stream and they did an episode on Qboard, I think two, three months back.
00:47:53
Speaker
And the one thing that I really liked was live migrations can also be defined through YAML files. Like I can have, is that correct? Like, yeah, okay. So I can have the VMs and then tell them, okay, this is when I want to live migrate. So I don't have to go to the UI. I can still keep using the Kubernetes constructs and my favorite YAML files to do live migrations.
00:48:13
Speaker
Right. And we sort of skated past the, you know, what happens when a Kubernetes worker node fails, like a physical host goes down. Right. So we've actually built technology into Kubernetes to do this, that normally Kubernetes takes like five minutes to figure out that a node's not responding. That's not quick enough for VM workloads, right? You got a database that people are waiting to get into, right?
00:48:40
Speaker
Now, the thing's down because the VM is not running anymore. We actually have an operator that will detect the node is down, it will fence it, make sure that it's powered off or disconnected from the cluster, and then restart those workloads elsewhere. Usually within 60 seconds, again, it's the same failover technology that we use on platforms like Rev and OpenStack just running on a Kubernetes platform.
00:49:08
Speaker
One other thing I did, we kind of skated around the storage thing too. Ryan, I think you mentioned this. This is something that's kind of cool and different and I think it's worth mentioning. Sure, please. So normally, you know, you as an admin would go build, I have golden images and I build them and you know, I got a process. I have a PowerShell or Python script that does this and it puts it in a data store and I do stuff with it.
00:49:34
Speaker
Sure, you could do that with Qvert, but we actually have something a little more clever, right? We actually can create virtual machine templates that have boot sources that exist in the cloud, right? So like rel, there's actually rel images that are in the cloud, qcout2 that you literally free and downloadable anywhere. Right.
00:49:55
Speaker
So you say, okay, that's what I want to use that for my boot source. And what will happen is we'll go create the template, we'll download the cloud image and make it available to the cluster. And you every VM that you build for that template, we'll use that latest image. Then there's a cron job that we create that runs in the background and checks, you know, to see if that image is updated. And if it is like, say you're running rel nine dot
00:50:20
Speaker
9.0 and 9.1 comes out, we'll go out and we'll grab that. ZStream comes out for a relevant. We'll grab the updated image, update the boot source, and then any new VM that you create from that template will automatically get the updated boot source without the operator doing anything.
00:50:43
Speaker
It's a self-service platform. Yeah. And so all of this catalog that you have built, does it have to be on customers, premises or customers? A bunch of clusters or only the images that they need can be pulled down as these boot volumes.
00:51:01
Speaker
You can do it either way, right? You can bring everything down and cash it locally, or if you've got a strong enough inter and a redundant enough internet connection, you can leave them up there. Our experience though is most customers, especially in the financial and government places, either use a proxy, because they don't allow their Kubernetes clusters to talk to the internet. Yeah, that's actually better that way.
00:51:31
Speaker
But OpenShift itself has the ability to create a local mirrored cache of everything that's up on the Internet. So you go create that.
00:51:41
Speaker
that's now disconnected from the internet and you can deploy anything based on that. Okay, gotcha. And since you hinted at customers, right, that brings me to my next question, like, are there any case studies or success stories that you can share where customers are using VMs on OpenShift in production and running this container plus VMs side by side on the same infrastructure stack?
Qvert Case Studies & Community Resources
00:52:05
Speaker
Oh, I think this is a very appropriate question because we have Red Hat Summit coming up. Yeah. And Red Hat Summit actually we have at least three sessions that are focused on OpenShift Virtualization. So Israeli Ministry of Defense is going to talk about the implementation of the private cloud with 20,000 plus active users. They needed a five nine sub time.
00:52:31
Speaker
And the reason they went here is to increase the liability through automated processes, component decoupling, and then implementing their governance. So a lot of stuff that Peter just mentioned to you, customer actually is going to talk about that. We have a customer on the edge from the National Oceanic and Atmospheric Administration. It's much easier to say no, but not everything.
00:53:00
Speaker
So they analyze weather data to help pilots know when it's safe to fly and, of course, get all these Twitter alerts and also they have like 150 cluster. They're moving them to open virtualization. They'll talk about that. Morgan Stanley has talked about previously about open virtualization. They'll continue to talk about where they are on this journey on the unified application platform.
00:53:27
Speaker
And then public reference on our website itself, we have a very interesting customer, which has an online classified site and called Sybenden. And people in the US may not have heard about them because they're based out of Turkey, but they have like 58.1 million monthly active users from the last time they reported when they did the case study.
00:53:52
Speaker
Yeah. And 12.7 billion page views, 5 million active listings. So it's kind of like an eBay price list combined. And they are like a 24-7 operation. So they moved their whole three-tier application onto the OpenShift virtualization. And now they're actually taking their front end. So the stuff that Peter was talking about, you take your front end and containerize it.
00:54:22
Speaker
it's actually happening. So that's actually one of the things which we did want to talk about a little bit is we talk about containerizing as the reason to go to OpenShift or to Kubernetes. That's not necessarily true, but it's also possible, right? So if you're doing change management,
00:54:45
Speaker
easy stuff maybe either to completely rewrite the application or lift and shift. So OpenShift virtualization, you can think of it as a lift and shift operation. You know it works, just get it in there. And if you are not feeling like doing anything more, it's still going to run at the VM. But if you really think you want to add more agility to your processes, like you want to change your
00:55:13
Speaker
Shopping cart on a periodic basis or something by adding containers in there. It's all free game at that time, right? So that's some of the advantages people get going this way. Absolutely. And I was going to say, since you mentioned Red Hat Summit, I know Bob and I will both be there. Will you both be there as well?
00:55:34
Speaker
Yeah, Peter is actually based in Boston, so he'll be there and I'll have another person. So you definitely catch up. And again, we are working with all our partners and both of you are in companies we love. So there we go. Yeah, we'll definitely I'll definitely be there. You know, it's what the beautiful Boston Seaport District. Yep.
00:55:59
Speaker
but it's gonna be long days and where they're actually OpenShift Commons is actually on the Monday. If you can find me there or we'll be all through the different sessions. There's actually a lab going on too. So this is the other thing that's important. When we talk to customers about this, they still, okay, yeah, you convinced me or those are beautiful slides, but I'm still not sure.
00:56:26
Speaker
The best way to do this is to get hands on with it right now. Okay, great. Now I need a bare metal Kubernetes cluster.
00:56:33
Speaker
No, not really. So there's actually something that Red Hat has built. It's called Q by example, has these different learning paths in it. And one of them is Qvert. So so again, everything that we do has got a strong upstream component and is freely available for anybody to use. So you can come up to Qvert and, you know,
00:56:57
Speaker
participate as much or as little as you want. The other the other contributors and adopters you'll see up there like Nvidia and you know other you know what people can see can consider competitors to Red Hat like SUSE, the rancher, the harvester folks are there.
00:57:19
Speaker
Absolutely. All right. Well, you know, I do have one more question before we can dive into sort of, you know, the community, you know, where people can find more and that kind of thing. But since you are coming at it from sort of a Red Hat perspective, and we talked a lot about Qvert and what it is, but, you know, maybe we can talk about the differences between what is, you know, Qvert and what is OpenShift VMs.
00:57:41
Speaker
Yeah, so it really is upstream project versus downstream product, right? The basic one is that Red Hat is doing all of the integration and testing of the full OpenShift stack, right? So it's not like, hey, I could go download this Kubernetes thing and now I got to go build it myself.
00:58:00
Speaker
And by the way, we have customers that do the do it yourself thing and they're actually quite good at it. But the thing you got to remember is it's not the building it that's the hard part. It's the keeping that running for three to five years and upgrading along the way. Red Hat takes that that part.
00:58:19
Speaker
So all the certification for rel and the Windows Virtual Machine, that's built into it. The one thing that you will find in OpenShift that is not going to be an upstream Qvert is OpenShift actually has a fairly capable and competent web console.
00:58:37
Speaker
It's actually a UI that you can manage all of the things in OpenShift that I just described, workloads, pods, network storage, and virtualization as well. A lot of the panels and UI that you're used to both in vSphere and Rev, we've actually replicated a good part of that, the useful parts of that in OpenShift.
00:59:03
Speaker
Believe it or not, I was using that today and I could really connect the dots when you were talking about certain features as I was going through it today. I'm still having some issues so I might reach out to you Peter. That's fine too. One of the other cool things about the web console is
00:59:24
Speaker
there's this dynamic plugin concept. So probably the best analogy loosely is sort of the vCenter plugin technology. So I know there's other companies and ecosystem partners of Red Hat that are building their own plugins that manage their storage and networking that plug directly into OpenShift.
00:59:47
Speaker
Got it. It makes sense. And speaking of getting help or contacting someone, I know we talked about us being at Red Hat Summit and you'll be involved there. I know you've mentioned Kube by example. Are there other communities? I know I've seen Kubevert.io and there's a community there. What other sort of resources and sort of communities or events can people find more information at?
01:00:07
Speaker
Sure. The Qvert.io is obviously the place to land. There's a couple of things there, right? You'll find obviously all the code and all the maintainers and stuff. But we have an annual Qvert Summit, right, which actually runs alongside of KubeCon. Actually, no.
01:00:26
Speaker
I take that back. It actually usually comes right just ahead of KubeCon, EU. So that's literally, it's like KubeCon, but it's all about VMs, right? And everybody like Nvidia shows up, hey, let me tell you about all the cool new GPU stuff we got going. There's a lot of performance and scale stuff that goes on too. So that's, you know, the Qvert Summit is, and those are all recorded too, so you can look at them later.
01:00:53
Speaker
KubeCon itself, which is a massive conference, there's always at least three, you know, KubeVirt focused things there. And it usually what it used to be like, hey, let's tell you how cool it is to run VMs and Kubernetes. And now it's really morphed into VMs and something else. Like, hey, let me tell you about virtual machines and service masher, virtual machines and pipelines.
01:01:17
Speaker
you know, using Tekton, right? It really is. If you're just talking about VMs and Kubernetes, that's kind of the boring part. That's the goal, right? Like we want to make infrastructure as boring as possible. Exactly, right? So the other resource I can think of is Kubernetes itself has a Slack channel that's open and there's a virtualization channel in there, which is where all the the Kuber folks hang out. And then the Ask an OpenShift admin.
01:01:46
Speaker
Yeah. So that that's, again, a Red Hat streaming property. But Andrew, you know, very graciously lets us come on there every couple of months and talk about what new cool stuff is is going on. And it's fun because you get to watch what's happening. And it's not just, hey, let me show you how to do this. But like stuff kind of often breaks in there. And and you're like, well, how do they troubleshoot this? And then you can you know, you can see us go through stuff.
01:02:15
Speaker
And then I think actually as part of Qvert.io, there is a there is a Qvert devel mailing list. I can't remember the exact name of it right now, but there's there's plenty of ways to get into it. Yeah, if you send us the link, we'll include it in the show notes. OK, we can do that.
01:02:31
Speaker
Yeah, we'll definitely do that. Well, um, I think that's about time we have today. I know we covered a lot of ground today. Um, so I appreciate you both, you know, coming out of Kubernetes Bites, talking about Qvert with us. I know, uh, it'd be great to see you next week. Um, well next week when this airs, it'll be this week. So it'll be good to see you at Red Hat Summit. I hope we do. And, um, just with that, I want to thank you for being on the show. Appreciate that.
01:02:53
Speaker
T is one last thing that, since we're under strict confidence here, everybody on the internet is going to be able to see this. What we've talked about today is, and what most people think about with Qvert is, hey, I run user workloads in a Kubernetes cluster.
01:03:12
Speaker
One of the other things that's been kind of challenging for OpenShift is running on different virtual platforms. And it can do that, right? You can run OpenShift on vSphere, OpenShift on Rev, OpenShift on OpenStack. We've got a capability that we actually announced in the last, what's new in OpenShift 4.13, which is hosted OpenShift clusters.
01:03:34
Speaker
And what this is, is you've got a bare metal cluster. And then using hosted control planes, you can actually spin up individual open shift clusters that run inside a virtual machine worker nodes. And now you can actually take a big fat server and slice it up. And basically, you know, clusters now become cattle themselves, right?
01:03:57
Speaker
and I can give this dev team their own OpenShift cluster and that dev team their own OpenShift cluster. It's a pretty powerful concept. It's in tech preview right now, but look for it to be GA as soon as we can make it so. Okay, so one follow-up question there, right? These OpenShift clusters that you are deploying on VMs, do they spin up their own control plane or is the control plane shared across all?
01:04:22
Speaker
It's each one has its own hosted control plane and hosting control planes are a whole, we do have a whole nother hour topic on that. But right now, what you need to do is, you know, if you run OpenShift on a virtual platform, you create three VMs and shove a control plane in it. Hosting control planes are
01:04:43
Speaker
those now run in pods. So they're much smaller and more efficient. And you can actually run them in different places, right? So I can actually take a central management cluster, run a multiple sets of hosted control planes on it that are all pod based and manage multiple node pools of bare metal workers alongside it. And the cool thing is, is that
01:05:07
Speaker
It's much faster. So spinning up an OpenShift cluster on bare metal, it's about an hour, 50 minutes. Spinning up a hosted cluster, 10 minutes. Okay, that's awesome. Not just for the amount of time that you will save, but also the resource utilization. That goes up because now Control Pin is running in pod. So thanks for sharing that.
01:05:30
Speaker
That's right. And the other benefit that is, you know, that multi-tenancy of I can give a dev team their own cluster and they can have their own RBAC access to it, right? I don't have to give them access to the bare metal cluster, but if they go in and they hork something up because they did something stupid, I'll just give them another cluster. Then that's not that that ever happens, no. It's never going to mess up anything that you give them. It's never.
01:05:56
Speaker
No, okay. No, I guess to continue Ryan's thoughts, right? Like thank you so much for being on the podcast. This was, this has been a great episode. Like we went into so many different kinds of details. So I'm sure our listeners would appreciate it as much as we do. So thank you. Thank you both. Thanks for having us.
01:06:12
Speaker
All right. Well, that was a fun conversation with Sasha and Peter. I know it's something that I feel like I've been really wanting to dive into more and get more hands on. I've used it a little bit, but not nearly as much as I want to. What did you get away from it?
01:06:27
Speaker
No, that was a good conversation, right? Like, again, we started with the premise that, oh, are people really using this? And then obviously at the end, they're like, this has enough features that people are using, like people are already using it and people who are not are evaluating it and will end up using it. But I think my one key takeaway was, like one of the many was,
01:06:48
Speaker
Organizations that are looking to modernize and start using Kubernetes don't have to worry about modernizing their whole application stack or whole estate. They can maybe modernize 10% of their applications right now, but they can still move or migrate everything over to Kubernetes or OpenShift.
01:07:03
Speaker
And then maybe gradually, maybe in the next two or three years, they can modernize the whole stack. But right now, this gives them that unified control plane. This gives them all the features that they were used to in their previous virtualization stacks. Many of them, Peter listed out during the pod as well, and allows organizations to take their time. They don't have to rush into these things and adopt technologies that they might not have expertise around. So that felt like a good solution to have.
01:07:31
Speaker
I think the second takeaway was like all the different features that peter spoke about right like the new ones that we gp operator from nvidia allowing gpu splicing i know in the ecosystem which again i used to work with a lot that was and those were like really important things that i have the admin virtualization admins needed like you couldn't dedicate an entire gp to a single the answer you.
01:07:54
Speaker
VMWare build that technology over yours and having that technology already built into OpenShift virtualization. I'm not sure about Q-word. I don't know if Peter highlighted that. But yeah, having that already built in, that really helps people ease, right? Like, ease their mind, like, okay.
01:08:10
Speaker
the features that I expect are already available so I can think about moving and not be worried about running these things in production. So those were some of the takeaways that I had, whatever you ran. Yeah, I think building on top of that a little bit, right? The part where we talked about not changing a workload if it's not broken, right?
01:08:27
Speaker
Having the ability, like you said, to kind of slowly move things over. It's not like you have to lift and shift everything you have in your VMware infrastructure or to open shift, or just Qvert and Kubernetes. But you can take something like an old .NET application that the guy's gone, I think Peter said. No one knows how to maintain it, but it's still working right. That's a perfect example. You can move it over, you get all these enterprise features like live migration and
01:08:54
Speaker
and failover and those kinds of things where, you know, you can keep things ticking along if you are modernizing that way or are moving, you know, your infrastructure and things like that. I think there's a really powerful piece of it. The other part, the key takeaway that I got is it's not like this technology is really that new. It's really been trusted, right? Cumu and Kubert has been sort of, you know, the backend technology, the pieces that make it tick have been around for a long time.
01:09:21
Speaker
running virtual machines on Linux and things like that. So it's not a net new technology that I think people have to put their faith in. It's more or less like it's enabled now as it always was in Linux just through the Kubernetes ecosystem. And I think that's a really powerful piece of it in sort of allowing yourself to kind of trust and use the workloads that way. It's kind of
01:09:45
Speaker
Linux has always been there. So it's really just the pieces are there and we connect your networks, you know, through CNI, we connect stores through CSI, we have all these pieces that we're just kind of plugging together. It's more involved in that 100%. I'm simplifying it way overly. But the point is, is that I think that whole piece of it makes I think the trust worthiness of it. And sort of the ecosystem really being able to move fast. I feel like this, this whole idea of running VMs on Kubernetes,
01:10:14
Speaker
was something I felt like I was talking about as something new, and now I feel like it's come to the point where it's like, no, we're just doing it, right? It's moved very fast, so that's really exciting. And I think the sessions that Sachin highlighted for the Red Summit, or the customer case studies like Morgan Stanley and Department of Defense in Israel, those guys are doing it in production today. So I'm excited.
01:10:38
Speaker
After recording the interview section, I think I went back to my session catalog. I scheduled those sessions. I want to see what your challenges were and how they overcame those challenges. So again, great stories, a great piece of technology. I love how it interoperates with Kubernetes already, like with CSI, CNI. So I don't have to change anything. I don't have to change my networking providers. I don't have to change my storage providers. Everything just works. Yeah, there you go.
01:11:02
Speaker
Exactly. Well, you know what? Well, maybe if we go to the sessions, we'll go snag them and see if they want to come on the show and talk about it. I would love that. All right. One thing we forgot to do, this Bob and I's fault, we forgot to do our chat at GBT question of the week. And we're just going to do it here live with us. And so the chat GBT question I asked basically for it to come up with a question based on sort of, you know, Kubernetes in a container is a fun one. We always do fun. We don't want to be boring.
01:11:30
Speaker
So the question that came up with us, if virtual machines and containers had a talent show, what unique skills would they showcase? And what do you, who do you think would win? Uh, okay. So I have not looked at the answer, right? My guess would be like, okay, I don't want to talk about their individual skills, but they were both popular bands on their own. Now you get to see like one show where they both performs. I think that I'm excited for that piece. I don't know what John G. Peter is fun.
01:11:59
Speaker
You're saying it's better together. Nobody wins. It's a tie. You're very fair. I'll have more fun if both of them are performing together. I don't have to buy two tickets. I don't have to go to two different venues. That's that's it for me. Give me a hockey game and a baseball game at the same time.
01:12:16
Speaker
But, you know, we're working on it. I just pictured the hockey rink inside of Fenway. Anyway, moving on. ChatGBT answered its own question and it says, in a talent show featuring virtual machines and containers, each would showcase their own unique skills and here are some possibilities. So it does go into unique skills.
01:12:34
Speaker
Virtual machines might exhibit their talent for isolation and security. They could demonstrate their ability to create complete and independent operating system environments, struggling multiple applications effortlessly. Their talent would lie in maintaining strong boundaries and ensuring the stability of each virtual instance.
01:12:52
Speaker
Pretty fair. Pretty fair assessment there. And it says for containers, containers on the other hand would flaunt their agility and efficiency and they could display their knack of rapid deployment, quick scaling and resource optimization. Their talent would be in seamlessly packaging application dependencies, allowing for easy portability across any environment. Also, you know, pretty valid answer.
01:13:16
Speaker
I'm visualizing like an opera show going on with like strict rules and like a conductor guiding everything and that's for VMs and then containers being the trapeze artists and just jumping trapeze artists. I like it. I like it. It does end with something here that says as for who would win the talent show, it's a tough call.
01:13:33
Speaker
It would depend on the criteria set by the judges and the preferences of the audience. No shit. But both virtual machines and containers have their strengths in your cases, making a delightful competition filled with innovation and technological prowess. So actually, it came to the same conclusion you did, which is like, they're both great, and it's going to be a great show.
01:13:57
Speaker
So there we go. Our talent show between virtual sheets, a container has been, uh, you know, hook line and sinker by Chad GPT. Um, okay. Last shout out to, um, again, to Reddit summit. This will come out during Reddit summit. And if you there, come find us, um, uh, otherwise that's all I had for today. How about you, Bobbin? Yeah, let's, let's wrap this. All right. That brings us to the end of today's episode. I'm Ryan and thanks for joining another episode of Kubernetes Bites.
01:14:30
Speaker
Thank you for listening to the Kubernetes Bites Podcast.