 All right, welcome to SuperUserTV. I'm Mark Collier, CEO of the OpenStack Foundation. Really excited to be back on SuperUserTV. And with me, I've got Alex Pulvy. And what do you do? I am the CEO and founder of CoreOS. Wow, you're kind of a big deal, it seems like. I heard you did a demo yesterday. Thank you, fire. Okay. Yeah, we did a demo yesterday. Yesterday we showed off some of the capabilities of how you can use these containers and container management systems, particularly Kubernetes, to deploy OpenStack itself. Now Kubernetes, isn't that like Greek or something? Yeah, I think so, I don't know. Yeah, it's Greek. Yeah, it's Greek. This is what happens when we don't rehearse. Just throwing language questions at you. So, okay, Kubernetes, what's so special about that? I mean, I hear people talking about it. What is it that's interesting about it? Sure, so Kubernetes is this way of managing server infrastructure that involves distributed systems and containers, typically pretty focused on microservices, although any application can run in Kubernetes. So it's a cluster management system, particularly meant for deploying containers. And it's built by the team at Google that built Google's version of this, which is known as Borg. Okay. And so. I mean, do they know anything about like scaling infrastructure or is this kind of a side thing? Google knows that things are a thing or two about scaling infrastructure. Okay, cool. So they're sharing some of that knowledge through this code called this Kubernetes thing. So where does OpenStack come in the picture? You did some demos yesterday, blew some mines, but like what? What was the demo? And like, I guess, you know, how does OpenStack and Kubernetes, how can they come together? Sure, so a lot of people think that these systems are at odds in some way when they actually solve different problems. Okay. So containers get compared to virtual machines a lot. What they should be compared to is like RPM or Dev. These are packages. Right, right. Containers are just packages of applications, okay? Then beyond that, the cluster management systems get compared to things like Nova or something when they should be compared to like Puppet or Ansible or Chef, okay? And the things that a container and Puppet and Ansible and Chef do for you are application lifecycle management, okay? Now Kubernetes at the end of the day is about managing applications, okay? And so OpenStack in this context is another application. Just like you can manage OpenStack with Ansible and Puppet, you can do that with Kubernetes and containers. Okay. The big difference between these systems is that in Kubernetes, the cluster decides what server to actually run the application on, okay? And Puppet and Ansible, the operator encodes that into those scripts themselves. Right, so it's more of a system that kind of understands that modern applications that are microservices oriented are going to live on multiple machines. Like the whole idea that like an app runs on one machine in one operating system is kind of a little bit of a relic. It seems like people, it still happens, it still works that way at some low level, but as you move up the stack and think about a whole cloud and a whole cloud app, you're always going to use like a lot of different servers. So you're sort of wanting to create a language so to speak or a system that, you know, speaks that language. As soon as many servers. Right. Exactly. Like any meaningful application deployed on a server is going to be on more than one server. Right. So to have a system that kind of embraces that natively is what it does. Cool, so yesterday, we called your demo, I don't think we even told you we're calling it inception, but we just put it on the calendar that way. That's not bad. For the heck of it, see how you would handle it. But so you had Kubernetes underneath, for lack of a better word, an open stack on top because open stack is really just another app or a series of microservices in a way. But then there's other people that we talk to like live person that, you know, they're running Kubernetes on top of open stack. So you have a world where you could end up having sort of like different layers and Kubernetes in different places. Is that, you think that's going to be common? Let's go back to Google because I think it's much easier to think about. So Google has their infrastructure as well. I get all my answers from Google. All container-based infrastructure for everything. Gmail runs in containers, but also, so does Google Cloud. Their infrastructure as a service, okay? So if you are a customer at Google Cloud deploying Kubernetes, that does not mean you have access to Borg, the underlying system that runs Gmail. Right. The different operations teams. One is the Google SREs that are on call for the production data centers at Google. The other is your operations team that is in charge of the Kubernetes cluster you just deployed on top of Google Cloud, okay? And so that's logically exactly what's happening with this under over thing. There's going to be a team that's responsible for your corporate infrastructure for running everything. One of the things that they run is the open stack cluster itself, okay? And then the customers of the open stack cluster are other business units or whatever that might not have access to the underlying platform at all. They don't even want it. The customers of open stack, they don't even know it's there, it doesn't matter. That's an irrelevant detail to them. They just talk to open stack and get a virtual machine and run their workload on it, okay? And so that's a perfectly valid use case. Now where it gets confusing for folks is in the open source world and smaller companies, it's typically the same people doing that base layer as the ones that are doing the VM provisioning on top of Nova and so on. And that's just where it gets confusing and our minds get muddled very quickly. But if you look at Google and the way that they have their container platform that runs all of Google and then they have Google Cloud Platform, the infrastructure as a service and those customers like Spotify and stuff and they don't have access to Google's raw infrastructure at all. They have access through Google Cloud. And that's the analogy. It's not just an analogy, it's actually a perfect example of what we're doing here. Yeah, and I think, you know, I was talking yesterday about like what's going to be the lamp stack of cloud and I think, you know, it's a good analogy to a point that has limits and I think what I really liked about your diagram you showed yesterday kind of showed the Google world is it's not really a stack, you know? It's more sort of service way of thinking about things. You know, they're living in a similar area where they rely on something underneath that's common but there, you know, it's not really, things are not really on top of each other. It's just like coexisting in an environment where, you know, it's some layer, I guess if you keep it in layers that there is some, it still works to a certain extent. You know, you've got the physical and then you virtualize it and you containerize it and then you have other systems on top. So there's sort of layers to it but really people are composing different things and so it's not really as simple as a stack in a way. Yeah, I think that there is maybe a data center stack a little, I mean that's what you're getting to. Linux is definitely part of that. I think that's like a clear winner, okay? And then there's, you know- I think this Linux thing is gonna be big. I think it's, yeah, it's really catching on. And then there's, you know, I think a container is gonna be part of it going forward. You need to package your application somehow and that's part of the stack. Every piece of the application needs to be packaged somehow, okay? And so a container is a generic way that seems to be quite popular for packaging that. And then beyond that, you need some form of cluster management system in this new world, a way of deploying and upgrading those containers. And that's where we think Kubernetes is gonna win on that piece of it. Okay. And so that stack- So you placed your bets. We placed our bets and then we packaged it up and make an enterprise ready with Tectonic which is literally that stack, you know? So like we're calling the thing Tectonic. That's what Tectonic is at the end of the day is that. All right, we gotta get your plug in here on SuperUserTV. I'm sorry, I didn't mean to. No, no. There isn't a name board at all. But so when customers go and take Linux and they take Kubernetes and so on, they're essentially building their own Tectonic. There's just no name for that stack at all. Sure, sure. Well, you might as well name it. We named it. Now you also have this thing called, what you call Giffy, I called Giffy. Yeah, for reasons that we're not gonna get into. But anyway, yes, so what is that about? I mean, it's another thing you're naming. Yeah, Giffy is Google's infrastructure for everyone. Okay. Model of running your data center that folks like Google and Facebook and Twitter do that is distinctly different than like how every other IT department does. Right. And that gets at the essence of what I was talking about where you have the cluster management team that's doing dynamic container-based scheduling and then you have the application teams on top of that that consume that infrastructure. Yeah, so talking with you and Craig McClecky yesterday, one of the things that I guess I didn't fully appreciate, but it makes sense now that I think about it is that, like you said, Google Compute Engine, their cloud service runs inside of, or on top of, or relies on Borg, which is essentially kind of this undercloud, if you will. It's an intermediate platform. Yeah, and so it's like they run their cloud in a cloud, they're in a way already. I mean, it's kind of abusing the word cloud a little bit, but that's kind of what it is, right? There's this massively parallel orchestrated system of containers on what, a million, two million servers. I don't know how many servers they have now, but I think it's well over a million. And then they said, well, with this great infrastructure, what can we do? We could build a cloud and offer services to people. So that kind of opened my mind a little bit to thinking about that in terms of why would OpenStack make sense on Kubernetes? Like, well, Google's cloud, which is not based on OpenStack, but they built some software to offer a cloud to end users, is running on something like Kubernetes, so why not try that for OpenStack? Exactly, I mean, it's literally giffy. We're doing it like Google's infrastructure for everyone and we're doing it with Kubernetes. We're doing it with Kubernetes and OpenStack instead of Borg and Google Cloud, which is all proprietary software that Google built. Right, right. Cool, well, I think open source is the way to go, so. I think open source is part of this new LAMP stack, for sure. I think so. It should be very popular. Yeah, yeah, you never know where it's gonna go, but I think we're seeing some good momentum. So I guess, what else should we talk about here before we wrap up? Austin, have you been having a good time at Austin? It is so great to be here in Austin. It's not bad. It's great. It's a cool town. It's a very cool town. That's my plug, you know, and if you vote for me for mayor, I promise free barbecue. Free barbecue for everybody. Yeah, I like to call your mayor.com. Minimum income for all Austinites is a barbecue platter once a week. I think that's a good platform. Yeah, it is a good platform. Yeah, which barbecue place would you pick? Well, I mean, everyone says Franklin's is the best and nobody wants to be a bandwagon person, but it's actually the best. That's the thing. I mean, it really is the best, but you know, but I gotta tell you who the sleeper favorite for me is opies. They're actually out in Spicewood. So, you know, people probably won't make it all the way out there, but it's opies out near Lake Travis and it's really incredible. Also, Terry Blacks is one of the better places that doesn't have a line, but it's just really good. Right. So anyway, thanks for joining us on barbecue talk here at Super User TV from Austin, Texas. Mark Collier and? Alex Polby. All right, thanks, man. Thank you.