 Live from Seattle, Washington, it's theCUBE on the ground, covering KubeCon 2016, brought to you by the Linux Foundation and Red Hat. Here's your host, John Furrier. Hello and welcome to theCUBE on the ground here in Seattle for a special presentation of KubeCon and CloudNativeCon. This is theCUBE's coverage. Our next guests are Josh Burkus with Red Hat and Perdipto Pachaharia, who's also with Red Hat. Guys, welcome to theCUBE, KubeCon and CloudNative. I mean, there's now events going on around this hot stuff. Thoughts on this. Obviously developer craze is all where it's all about. We just talked to a hot startup in Seattle that was lucky enough to come in and talk to us. It's truly changing the game right now and the developers are driving all the action. What's the conversations like? Obviously orchestration, Kubernetes has more job openings and I think then I've seen in a long time it's like straight up. There's so much action in Kubernetes and containers. Well, it's as, see, I've actually been on the Ops side. I've been in Ops software geek for 20 years. This is actually very exciting for me because if you had told me even like six or seven years ago that we'd get, what do we got here? 2000 people at a conference to talk about automation and distributed systems. I'd have laughed at you. Be like, oh, you get like 100 guys at Berkeley to talk about that. Not like large groups of developers. So it's very exciting that we're actually like, trying to move forward the state of the art of software. Not just what we do, but how we do things. Pradipto, you know, one of the things we were talking about earlier and all in the queue all the time is we've been all big fans of DevOps going back since we started SiliconANGLE. But DevOps now is so mainstream and actually have a name for it. It's called Cloud Native. So Cloud Native has kind of become the DevOps word for just how things get done in software. Right, I think it's the generic software thing that you come up with new term to take it forward. But it's just amazing how people started from Agile then they go to Cloud Native. They went to the DevOps whole revolution and now they call it Cloud Native. It was amazing that they just announced the next conference right now, which is a year from now. And it will be apparently two times bigger than this or three times bigger than this. And they already announced the Berlin one. So it's obviously it's like global thing across everybody, across everywhere where we need so many people and so many people are obviously attending the conference. It's like full house there. I had to take a whole panorama picture of all the people in the whole room. I mean, to me, it's so exciting for me that I was one being a software guy and seeing this kind of celebration of the tech. So it kind of makes me kind of get a tear in my eye. But more importantly, as an entrepreneur, there's huge wealth creation and value creation around software right now. You're seeing it as a global phenomenon and the invention of this unknown things are happening in front of us and the developers are on the front lines doing it. And that's a sea change. Now, that's all great. I guess everyone excited, but you got to operationalize it. This is where the challenges are. So, Josh, talk about the things you're working on. Project Atomic is one. This is a big deal, this operationalizing, this cloud layer thoughts. So, and a lot of it is, we're looking at how do you actually do en masse things that people didn't at Boutique way before, right? 10 years ago, you wanted to actually scale out a system and make it automated and make it a real distributed system. You hired a bunch of Brainiacs from a university and you put them to develop something custom just for you as opposed to actually making it productized and really at every layer. So, one of the things we're working on for Project Atomic is when you start looking at these container-based orchestrated systems, what you need out of the operating system changes, right? Because you're not installing individual operating systems and then configuring them by hand anymore. And configuration management kind of brought us towards that but not really where we want to be. So now we're looking at, hey, if I'm looking at an individual host, not as an individual host, but as part of a whole group, how do I want to manage that? And that's what Atomic is about, is that we want to actually manage the individual hosts en masse the same way that we manage the containers. That's Atomic host. That's Atomic host, yeah. And what's your goals for this year? I mean, obviously it's a lot of community rallying and going on, what's your goals? So, our goals is to actually, well, two things. One is to add the rest of the features that we need on the server. And we just recently made a milestone in that so for both the RHEL Atomic host 7.3 and for Centos Atomic host and Fedora, the open source versions, we added layering so that you can actually deploy because Atomic host, it's an imaged OS that is the OS gets installed from an upstream image from a server so that you can distribute en masse rather than configuring on site, right? But you need to make changes to groups of servers. And so we added a layering feature so that you can layer extra things on top of that image in the same way that you would inside a container, but this time in the host OS so that you can actually have groups of servers that have special purposes because they need extra software like storage software or one of the big things I've been using it for is to install different versions of Kubernetes because I also mess around a lot with Kubernetes, right? And so I want to have 1.2 in the server and 1.4 in the server and 1.5 alpha in this other server. So it allows you to focus in on things you want to get done. And how is this reinventing the OS because this is a shift in thinking. Right, well, the thing is that basically the way that we have distributed managed software for the operating system really hasn't changed in any fundamental way in almost 20 years, right? It's been packages, right? RPMs or Debs or whatever, getting distributed to stuff, getting installed and then getting configured, right? And we added some configuration management here, but we didn't really change how we actually do things. And so moving to an OS which is based on an image tree really changes that. And also with the idea that the base OS should be lean and that the stuff you're installing on top of it should be installed in namespaces, containers and on the desktop, there's a project called Flatpak from the GNOME for installing desktop stuff. So one of the other things that's coming out of this is because this came out of Red Hat because Atomic came out of Red Hat, it's all been very focused on the server. But in the meantime, a lot of our users in the open source space particularly in the Fedora project are like, well, I want to use this for desktop because like I want to run the latest cutting edge version of Fedora but I want to be able to roll back and recover if it stops working. Yeah, what about security too? This is another concern, security. Yeah. Yeah, and that comes from sort of a different place. I mean, obviously big enterprises are now looking at automating in this way. And a lot of them had actually gotten very used to the level of isolation that virtual machine based infrastructure gave them or real machine based infrastructure gave them. And containers don't deliver the same level of isolation. And so we need to rely a lot more on security tools built into the OS and built into our container stream. That doesn't disrupt Project Atomic. It facilitates it, doesn't it? It facilitates it, yes. Yeah, because we're looking at building in all of these sort of tools, right? And so Project Atomic is sort of an umbrella and like one of the tools we have under there is Bubble Wrap, which is a way to actually run non-root containers even under Docker which has some issues with running non-root containers. So that you can limit people's permissions. We do a lot of work with SE Linux because the Red Hat platform is devoted to SE Linux in order to do that. And lately we've been doing a lot of work on image identity and image fingerprinting to know that you're getting the image that you think that you're getting and that it's been verified. And that's a big deal actually in the whole container world but it's very important, right? We have a tool chain for RPMs and Debs and regular packages for that. We need a new tool chain for containers. So real evolution and tool chain. Pradeep, to talk about open compose and what you're working on specifically at what stage is this project? What's the update? Sure. It's a very new thing. It was literally released like the initial cut or the POC version was released like last week. So the idea is Kubernetes is awesome and but it needs sometimes you have to do a lot of work to get it. So a higher level abstraction is always a nice thing to have at many times which will help developers get to use Kubernetes or deploy their applications on Kubernetes or the microservices very easily. As a result, we thought about it how it can be done. Can we create something abstraction that can be fed into tooling maybe and then deploy the applications? That's how the idea came into picture first of all. So what we have done is we have created a back for now and a tool and we have been experimenting with it internally with some various microservices that we are building internally and we're trying to deploy them on Kubernetes or OpenShift and it works great for now but we would like other people from the Kubernetes ecosystem and other orchestrator maybe even if it can be made. Composability is what everyone wants. They want to have it basically easily to compose apps and who's involved in this right now? So you know Red Hat's behind it. What causes it? We are talking to Google already. We are talking to, we have our initial conversations with CoreOS as well and then there is a smaller company called as Skipbox. So we are talking to a bunch of people who are whomever can be. We would ideally love to talk to MISOS and the MISOS ecosystem as well if it makes sense but the whole idea is as you said composability of services and tooling around. What's the main attraction to get people on board with this? High level abstraction, make it easier. The user experience should be really, really simple and easy to use and it should be very simple to teach anybody to get on board on this. And where's some information that can get GitHub and code out there? Yes. Give us a quick word on that. So there is a repository called as Open Compose inside the Red Hat Developer Organization. That's where it resides. Now the spec resides there. The tooling and basically the read me should tell you everything. It has a link to the toolings and everything as well. Plus, we have written a detailed blog post including some code examples on the, we can share that link with you. So that actually can take you to the details of the project, yes. Josh, final question. How does Atomic, OpenShift and Kubernetes look like? How does it all come together? Okay. So Atomic is the underlying layer for whatever you're going to run as your orchestration system and your application management system on top of it. The idea is to actually provide a really solid base for that. So like when Red Hat, if you're doing OpenShift online and you're getting supported OpenShift from Red Hat, it's running on top of Red Hat Atomic. But say if you don't want to use OpenShift, you have a different tool chain, you want to use Kubernetes, you know, mainstream Kubernetes for that without the addition of the OpenShift tools, then we're going to actually run that. And like one of the reasons for adding this new layering feature was that you could run whatever version of Kubernetes you wanted. I've got a microcluster down in the booth running 1.5 alpha right now. Guys, thanks so much, OpenCompose, Atomic. Kubernetes, just quick, give you guys the last word. How hot is this area? I mean, give some color, personal opinion, pontificate, opine, share your thoughts on what's going on. It's big. I mean, like I said, in the future, everybody doing just about anything is going to be doing a distributed system and Kubernetes gives them the tools that regular ops people, regular devs can actually do distributed systems and have them work. I would call it democratizing the cloud, I think. It makes it easier. You don't need to be the sysops guy who knows everything. It makes it like... It's an operating systems guy's dream. That's supposed to my degree and undergraduate. And I love the cloud as an operating system. If you think of it that way, the world is your oyster, as we say in Seattle. Thanks so much for coming to Kubernetes. We're in Seattle. We have the best oysters and the cloud is one big operating system. I'm John Furrier, you're watching theCUBE on the ground. Thanks for watching.