 From Seattle, Washington, it's theCUBE, covering KubeCon and CloudNativeCon North America 2018. Brought to you by Red Hat, the CloudNative Computing Foundation and its ecosystem partners. Okay, welcome back everyone live here in Seattle for theCUBE's coverage of KubeCon and CloudNativeCon 2018. I'm John Furrier of theCUBE with Stu Miniman breaking down all the action. Three days of coverage, we're in day two, a lot of action at open source. 8,000 attendees, up from 4,000 North America. They're in China, they're all over Europe. The community is growing in a massive way. We have two great guests here from Red Hat, all making it happen. Part of the community we've got Dan Mueller, who's the Kube alumni director of community development many times on theCUBE. Good to see you and Rob Somsky, Somsky principal product manager, both at Red Hat. Guys, thanks for coming on. Great to see you again. Yeah, glad to be here. So the world's changing a lot and there was some news recently around Red Hat. I can't remember what it was recently, something big news but you guys have been big players in open source for years. We could always cover it. We always wax on about the origination of it and how the evolution but the CloudNative piece has gotten so real and you're rolling in particular. We've had many conversations going back to the open stack days of how OpenShift was developing then the bet on Kubernetes and you guys made CoreOS acquisition. Those two things I think to me, at least from my perspective, really catalyzed a lot of things at the right time. So from there, a lot of things has just been happening really in a good way. Big tailwind for you guys, CloudNative app developers are using open source, CI CD pipeline and then also policy based up under the hood. Completely big shift in moving the game down the field. So congratulations, first of all. But what's new? What's the update? The update is operators. I think the next big thing that we're really focusing on and that's the game changer for all the second day operations type things and we'll make Rob talk about it in detail is the rise of Kubernetes operators. It's not a scary thing. It's not like Terminator Day or anything like that. But it is really the thing that helps us make the service catalogs, the Kubernetes marketplaces really accessible to all of the databases as a service and all of the other things and takes out some of the complexity of delivering applications and databases as a service to anybody running Kubernetes anywhere. Take a minute to explain Operator real quick and then we can jump into it because I think this is a fundamental trend that we're seeing. Developer trend is pretty obvious. Been that for a while. CloudScale, ML, machine learning, all the goodness around application. But the operator side of it has been an IT thing. But now you guys have a different, a new approach that's winning. What is it? What is Operator? What's Kubernetes that has the approach? And I'll let you... Yeah, so it's basically like the rise of containers is great because you can take a single container and package an application and give it to somebody and know that they can run it successfully. And Operator does that for a distributed system in the exact same way. So you're using all the Kubernetes primitives so you're not reinventing service discovery and secret management and all that. You can give somebody an entire Kafka stack or a machine learning stack or whatever it is that these very complex distributed systems and have them run it without having to be an expert. They need to know Kafka at a high level but not exactly all the underpinnings of it because that's all baked into software. And the benefit and the impact of the organization is what? And just to clarify this, so this was added in, I believe Kubernetes like 1.7. It's something that's in there. It's not something Red Hat specific. Yeah, it's got, yeah. So you're extending Kubernetes. So you have a custom resource definition which is an extensible mechanism for saying, hey, I've got a deployment or a staple set but what if I want to have a new object called a MongoDB that knows how to deploy and manage and upgrade MongoDB? So that's the extension mechanism that we're using. Yeah, so you've got to think, there's certain applications that this is going to make just a lot easier, how I manage them, deploy them, things like that. Any specific examples you want to share as to? All the cluster databases? I've just seen that there's a lot of, the application side in this model have been very excited about this. Yeah. So it's all the vendors and partners that want a hybrid cloud story just targeting Kubernetes. So we're using Kubernetes under the hood and then everybody wants to run a staple database tier whether that's Mongo and Couchbase, Cassandra, whatever. And these are all distributed systems. All right, so I want you to just parse. You said a hybrid cloud, explain that model because that's something, in general, discussion is hybrid or multi means I'm running multiple places, I'm not necessarily stretching an application but I have instances there, just want to make sure I'm on the same page. So this would be more the compatibility that you're programming against when you're building an operator is Kubernetes. It's not a cloud offering, it's not OpenShift, you're just targeting Kubernetes. And so you can run MongoDB on-prem in the cloud and have it function the exact same by standing up one of these operators. And then if that operator has higher level constructs for how to do multi cluster aware data rebalancing, you can take advantage of that too. And the open source status of this product is what? It's all open source, it's all in the GitHub repos. There's a Google group for operator framework that anyone can come and participate in. We hold SIG meetings on the third Friday of every month, 9 a.m. Pacific time, and it's a completely open source project. There's a whole framework around it. So there's the operator SDK, the operator lifecycle management and operator metering, all the tooling there to help people build and manage these operators and it's all being built out there in the open with the community support and feedback loops. What's the feedback? What's the top feedback you guys are getting right now? I have to say, this is really like, I've been hanging out with you guys like the past three, four months on this topic, trying to get my head around it and everything. And we came here and we had two sessions, an intro session and a deep dive session, intro yesterday, a deep dive today. Today's deep dive, the room was about 250 people and there were people outside of it. Security guards, blocking people from coming in. Nobody could come in and it's insane. It's like, everybody needs these things and everybody wants to figure out that. And when you ask people in the room who's building one, half the room raises their hands. You know, it's just, it's great. This thing crept up on us really, maybe not on CoreOS, okay? Crept up on me very quickly. So, and it's very rapid adoption. We have a Kubernetes operators workshop on Friday. So not only do we have pre-conference days of like OpenShift Commons and stuff like that that are huge now, but now we're starting to bookend CNCF events and put on other things just because, and that, we had 100 seats that we thought we, you know, we were hoping we would fill and it sold out like in minutes once it got in there and there's a waiting list of 300 people. It is like one of, you know, aside from Knative and all the other wonderful hot things too and how, it is one of the most interesting developments I think right now. Thirst for the content. What an impact. Yeah. You can get all of, all of the documentation is out there now and people are already building them. We have a list of 50 community operators. It's just, you know, it's phenomenal how quickly it's growing. You know, Diane and Rob, it's funny because you know, we do so many of these Kubernetes we're doing for this our 10th year doing theCUBE coming up and I remember the conversation going back in the OpenStack days. We would ask questions like, if you had a magic wand, what would you like hope to have happened, right? And you know, those are parts of the evolution where it's like, it's aspirational, things are being built. It seems now with Kubernetes, it's like, I mean, it's almost like, wait a minute, it's actually, this is actually, the goodness is so compelling. Yep. Above and below Kubernetes that it's almost like, uncomprehensible to think about, okay, this is actually happening. Finally, the kinds of steady state kind of operational things that have been a pain in the butt for the years. The material, it's gone. Yeah. For the most part. Yeah. So, Rob, I've been having a lot of, just thinking back to, you're employee number two at CoreOS. When I first talked to CoreOS, it was, you know, we're going to build all of these, you know, individual tools and we're going to open towards them and it's going to be good. We've watched this just rising ecosystem in the CNCF and feels like what's nice and what's different that I see compared to some previous things is it's not, you know, one product or even, you know, small group of companies. It's, you know, I have this toolkit and some of them work together, but many of them are independently used. You know, we talked to your peers earlier about LCD. LCD is totally standalone. Doesn't need to be Kubernetes. You know, what have you seen, you know, if you go back to that original vision with, you know, CoreOS just been, you know, part of this whole ecosystem and done it, if this was available and is this delivering on the promise that, you know, your team had hoped to work on. Yeah. So we've always filled in where we see gaps and so something like LCD, the concept is not new. It comes from Google and they have a system internally. And as Brandon got up on stage and said, we needed that to coordinate reboots across a cluster of machines. Didn't exist, so we had to build it. Same thing with how we wanted to manage Linux. There was no distro that even really resembled what we were doing. Wanted to do automatic upgrades. People thought that was crazy, so we had to go build it. And so, but we always adopted the best debris technology when it existed. And our early bet on Kubernetes, we just saw this is the thing and went for it, you know, I don't even remember what version, but it was months and months and months before 0.0 or 1.0. So it was, we've been doing it forever. And you just see the right thing and it's the little nugget that you need. And if you don't see it, then you build it. What are you surprised about, Rob, in terms of the ecosystem now? What you mentioned, some goodness is happening. Still a lot more work to do. Visibility around value creation. You start to see spots where value can be created in the ecosystem, which is great. Still more work areas. But what's surprising you? What do you see as opportunities, challenges, your thoughts? Because this vision of ease of use and programmability is happening, right? So it's still more work to do. What do you, what's your vision there? What's your thoughts? I mean, I think self-service is key. So this is like the rise of the cloud comes from self-service for developers. And Kubernetes gives you that right abstraction where self-service for VMs like OpenStack, which is not quite at the level of what you wanted. You didn't want a VM. You actually wanted a place to deploy an application. You wanted load balancing. You wanted service discovery. You didn't want like a bear, a bunch of VM. And so Kubernetes raises you up to where you're productive. And then it's about building stuff on top. But what's interesting in the space is we're still kind of competing on Kubernetes installers and stuff like that. So we're not even really into like the phase where people are being super productive on the platform other than these leading companies. So I think we'll democratize that and we'll have a whole new landscape. And so 2019, you see as what being a key theme for Kubernetes? I think it'll be core stuff built on top like all the serverless frameworks, a bunch of container native storage solutions, solving some of these problems that folks are reaching out to external machine learning but bringing that on to the cluster, GPU support, that type of stuff. It's all about the workloads. And traditional end users, you guys have a huge install base with Red Hat being well-documented. As the end users start coming in and looking at cloud native and doing the reimagining of their environment whether it's IT spend, IT investments to how they run their coding and their deployments, it's going to change. 2019 is going to have an impact that I call mainstream enterprise for lack of a better description. What's the impact of those guys? Now they now have headroom. They can do more. What's the mainstream enterprise look like right now with Kubernetes, the impact of Kubernetes? I think they're going to start deploying applications and get like lower the time to business value much, much lower. And you know, I was just talking to a customer, they ordered bare metal machines like a year ago and they're still not raked in in the data center. And so the people are still getting over that type of stuff. But once you have like a shared Kubernetes layer, you can onboard teams like crazy. I mean, namespaces are free quote unquote. So you can get 35 engineering teams on a Kubernetes cluster, super easy. So they can ramp up a development teams basically as they bring value in-house for outsourcing everything as they start getting development teams this is where the action is. I think you're also going to see the rise of those end users contributing back things to the Kubernetes community. And you know, as Lyft and Uber and everybody are great examples of that. You know, Uber with Jaeger and Lyft is we're just in the operators thing and you know, they raise their hand that they're about to open source it, you know, a few operators that they're building and stuff. And you're just going to see people that you didn't normally see. You know, often these large foundation driven things are vendor driven. But I think what we see here is the end user community is now embracing open source is getting the legal teams there, allowing them to share their things because you know, one, they get more people to maintain them, you know, and more people working on them. But it's, you know, it's really, I think the rise of the end user we'll see as they start participating more and more in here and you know, that's what, that's the promise of open source. And that's where CNCF really made its bones. It wasn't really vendor led per se. It was really end users driving the guys building out their stuff for the first time. You see Lyft for instance, great example. You guys did a CoroS and like, this is like the new generational model. Yep. Final question before we break. I want to get this out there. Get a plug in for Red Hat. What do you guys, what's the focus for the show? What's the news? What's the big story for Red Hat here at KubeCon this year? I think it's operators is what we're here talking about. It's a really big push to once again, get smarter workloads onto the cluster. We've got a really great hybrid story. We've got a really great over the air upgrade story that we're bringing from some of the CoroS technology. And then the next thing is, you know, once it's easy to run 35 clusters, we need a bunch of workloads to put on there. And so we want to save folks from the toil of running all those workloads as well, just like we did at the cluster level. Awesome, awesome. Well put, I couldn't add more. I think it's definitely, one of the things that CoroS did, you hit the nail on the head earlier, is when there was something missing, they helped us build it. And with the operator SDK and the lifecycle management and the metering and whatever else the tooling is, they've really been inspirational inside of Red Hat. And so, you know, they've filled a number of gaps and it's just been all operators all the time right now. It's great when a plan comes together, you guys got a great tailwind. Congratulations on all the success. And it's just the beginning of the wave. It's theCUBE covering the wave of innovation here at KubeCon CloudNativeCon 2018. But back with more live coverage, day two of three days of CUBE coverage. We'll be right back.