 Live from Vancouver, Canada, it's theCUBE at OpenStack Summit Vancouver 2015. Brought to you by headline sponsors EMC and Joyfully by Red Hat and Cisco with additional sponsorship by Brocade and HP. And now your hosts, John Furrier and Stu Miniman. Okay, welcome back everyone. We are live here at the OpenStack Summit in Vancouver, British Columbia. This is Silicon Angles theCUBE, it's our flagship program. We go out to the events and extract the signal from the noise. We'll talk to the hot startups, the CEOs, the venture capitalists, the big companies, and of course the developers and share that with you. I'm John Furrier with my co-host, Stu Miniman. Our next guest is Patrick Riley, the CEO of Kismatic, the Kubernetes company as they say on their website. Patrick, welcome to theCUBE. Thanks, it's great to be here. Great to see you here, great to have you on to share some insight around Kubernetes, what's going on with that. I'll see the background of the story. But before that, tell us a little about your background, the company, how it all got started. What's going on? Yeah, so I've worked in highly trafficked websites for quite a while. I sold my previous company, Orly Atomics, to the folks at Mesosphere, which really got me thinking about infrastructure again and how we can make it better for everyone else. And at Mesosphere, we worked closely with Google on Kubernetes from the beginning. And I just couldn't believe there wasn't a standalone Kubernetes company that wasn't Google. And it seemed like a great opportunity. So my co-founder, B.C. Burchard, and I decided to leave Mesosphere and start Cosmatic, the Kubernetes company. Talk about Kubernetes. John, I'm sorry, if I could just give some baseline because I tell you, a lot of people are still relatively new to containers. They hear kind of containers and Docker and Rocket and CoreOS, Mesosphere and Kubernetes. Can you just kind of lay out how they fit into the container discussion and how do those two go together, Kubernetes and Mesosphere? Sure, yeah. So with the recent hype of containers and because of the good work that Docker has done, people need a way to orchestrate those containers. It's one thing to get your services running in a container. It's another thing to figure out how you're going to actually orchestrate that and do things like service discovery. So with Mesos, you're able to treat your whole entire cluster of machines like a single compute resource and schedule Docker containers to come up and down and set up service discovery and use something like Marathon or Aurora to manage that cluster. With Kubernetes, it's a very similar approach to it and you're orchestrating those containers. But some of the things that are opinionated, so you don't have to worry as much about service discovery and you can trust that it's done a sane, default way. So it really kind of changes everything for people. There's a lot less tooling that you need compared to the current way of doing things. So talk about the Kubernetes rage right now because obviously containers has gotten everyone's attention. Obviously containers have been around for a while in computer science. It's not a new concept, but the timing of it was pretty awesome. And Docker's done some great work. How did Kubernetes come out and where did that traction come from? And you mentioned you saw an opportunity, but why is it relevant? Why is it dominating the discussion? What are the key elements of the problem statement does Kubernetes address? Well, I think if I step all the way back to before Mesos existed, you know, you had Ben Hinman, the co-creator of Mesos, he did internship at Google while he was pursuing his PhD. And he got really interested in the Omega white paper that came out. And he's like, basically I can build one of these. And so you take a decade's worth of learnings from Google that they put out in white papers and you have people so excited to try to implement these things themselves, then to have the actual Google themselves say, we're going to do a green field effort. We're going to take everything we learned from Borg and Omega and we're going to start from scratch. We're going to do it in Go. It's a much easier, you know, Mesos was written in C++. It's difficult to find good C++ talent out there. But I can take a recent CS grad, teach him Go and actually have him being a productive member, adding features and plugins and things to Kubernetes. So I think that's the key component. The fact that Google is saying, we want to do this, we want to open source this, we want to get it in front of everyone. And it's not just Google doing it, it's the same engineers that built Borg and Omega. I mean, you've got John Wilkes, you've got Brendan Burns, Joe Beda, all of these guys that have been working for 10 plus years at Google, coming in and saying, this is what we're going to do. And Google currently launches about two billion containers a week. Most of their workload is in containers. I'm going to trust them more than I'm going to. They have track record at scale, massive scale. Yes. So talk about the Google dynamic because we were commenting with the Google folks at their cloud platform conference about Kubernetes almost didn't get open source but Google's been involved in a lot of big megatrends. MapReduce, Hadoop, I mean, a lot of their white papers have rendered themselves into industry leading unicorns as they say. So you look at Cloudera, took advantage of the Hadoop market space as an early leader. You guys are enabling now with the open source component, although now Kubernetes is open source, you guys built a company around that. That's a great opportunity. Talk about the dynamic between you guys, Kismatic and Google relationship, how you guys work together and whatnot. Yeah. So the thing that we're trying to be as a company is the pure play open source Kubernetes much in the way that Mirantis is the pure play open stack company. So I don't actually want to try to make money off the open source at all. I want people to come to us because they want thought leadership. They need architecture reviews. They want to do paid POCs. They want to really just fast track their ability to get up to speed on this and get a best of breed approach. So we find working with the Google Cloud Platform Team to be great. They provide a lot of process guidance, a lot of thinking that we have a weekly meeting on Fridays where the whole community comes together and everyone kind of syncs up and it's exciting. Yeah, so it's interesting because one of the things people kind of criticize Google a little bit on is they do so many projects out there. And by the way, what you see out in the public is what they were using last year. They're on to the next thing. We talked about Omega and Borg and everything like that. We look at Hadoop. They're doing the next thing. So that kind of almost a void that you guys are filling is that we don't expect, I mean Google's done a ton with Kubernetes obviously, but they're not going to help all the enterprises. They need partners and others in the ecosystem to help fill that gap. Yeah, and I think when you talk to people like Brian Stevens, the SVP of Google Cloud, like that's really his approach. He can't put a warm embrace around one company. He wants to build a rich ecosystem, Kubernetes community, and getting people participating. So every company out there that's talking about Kubernetes here at the OpenStack Summit gets me excited because they're all coming at it from a different angle. And it is one of the first times that Google's willing to do things in the open. The code's all on GitHub. It's being developed there. There's no secret private repo where things happen behind the scenes. And it's an amazing opportunity where I can go to GitHub. And if I take the time to write a good specification, there's a high likelihood that I'm going to get one of the engineers at Red Hat or Google to actually do the work. And that's very rare. So you did an interview a couple of weeks ago. I saw the blog post on Morantis' blog. Mentioned that you guys are readying an OpenStack distribution of Kubernetes. What's the status of that? Any updates on that? And how does that relate to this whole container as a service? Is it different? Is it the same? What's this all about? Yeah, so our focus with Morantis is on the Morano application for installing Kubernetes. So right now today, if you wanted to have a Kubernetes cluster for development, you could order, well, actually, it's not going to be available for a few months, but the appliance, the unlocked appliance that Morantis is showing off with their partnership of Redap, you could order that device with two clicks. You could have Kubernetes up and running. And from the developer's perspective, they've got Kubernetes, they can start using container registry, get things up and going locally. From the ops perspective, they got the richness of OpenStack, the support tools they have there. And it's a pretty holistic option for people. So we're excited about that, and the Morantis folks are. You feel good about that? Yeah. Is there anyone testing? Is it beta? What's the code base look like right now? So the code is publicly available in the app store that was launched yesterday. You can go ahead and pull down the Kubernetes app if you have Morano installed in your OpenStack cluster. Now is Morano their stuff? Morano is actually the OpenStack foundation. That's the package. Yeah, so it's the packaging product. That's the buzzer on this container as a service thing. So there's some other plays in the OpenStack community right now with Magnum, but Magnum is based on Kubernetes as well. So the idea is we just want to get you to Kubernetes in whatever way that we can. How do you explain Kubernetes to the folks out there that are curious, they are interested, they've heard good buzz from their peer groups, they see some action in the ecosystem. What is Kubernetes? Why should, if you be something on their top of mind, why should someone even pay attention to Kubernetes? What do you see as the relevance and importance of Kubernetes? Yeah, so we already touched on the origins and coming from Google and why that's important. And I think if people want to move to containers, if they're comfortable with containers, and they need a way to orchestrate them, it makes a ton of sense to use Kubernetes. You have a rich CLI. We actually had Cosmatic worked on the building the web UI for Kubernetes. So you can have a nice web UI to look at and have feature parity with the CLI commands. And it's logically thought out. I can take someone and run them through a quick start on Kubernetes and it makes sense. There's not really any confusing concepts. You're up and running quickly. You can scale up and scale down, things like rolling updates. Everything's pretty sane approach. I've not shown anyone yet and they're like, oh, that's ridiculous, I never would do that or that's not the way I think it should work. All right, so Patrick, on the flip side, what isn't Kubernetes good for today? What work needs to be done to mature it? The keynote yesterday, Mark Collier said, it was Marano, Kola, and Magnum all early and the whole container space is really early. We're not sure who's going to win. We know it's hugely important. So what's the hard work left to do and what should we be looking for from you and the ecosystem? Well, I can answer that in two ways. On the Kubernetes side, we need to get to stateful services. We need to have those primitives built in so that we can be able to run things like databases and do it in a sane way. And it's much the same built that Mesos is in. They're adding those primitives as well. Once we get those, we'll be able to do different types of applications than we can today. I think another bottleneck is just the overall scalability. For the V1 version of Kubernetes in July, we're looking at having 100 nodes, so 100 physical machines and 30 pods per node as kind of our benchmark. We want to make sure everything works perfectly at that level. Obviously, a lot of companies are going to need to go past that. So when we get to version 1.2, 1.3, we're going to slowly ratchet up those numbers and try to get to actual Google scale with Kubernetes. And just really quickly back to the container piece, I'm really looking to companies like Intel to drive the ball forward there and make sure that they use their Linux kernel developers to add the namespacing to the Linux kernel to add the container features right into the kernel, do things like the Linux distribution that they just announced this week at the OpenStack Summit and get that to drive those features in the kernel. And I think it was interesting to me to see with their effort around their clear container stuff, they had to actually make kernel updates and they had to upstream those updates. And that just shows you how dependent we are on the underlying systems and with those types of things, you don't typically have people willing to try new kernels as they come out, right? Yeah. So talk about the funding and where you guys at the company. You guys take outside funding, who's involved, what's your goals for the company next year, hiring, give us a quick update on Kismatic. Yeah, so as we see Kubernetes maturing, we really kind of shifted our focus to be more service and support oriented. So that also kind of shifted our funding. So right now we're working with $3 million bridge financing while we try to figure out what partners to work with in a service and support capacity. As you might imagine, most Sandhill VCs aren't really keen on hearing service and support company because that's typically very hard to scale on people, they want to scale on product. So a few of the institutional investors that we're working with and some strategic investors, they understand why we want to do this and the way I present it to them is typically that we have an expiration date of this. We're trying to make sure that we get Kubernetes into clients, we get companies using Kubernetes effectively and then we can figure out what are the products that we're going to add on top of there. So our key focus for products right now are just that last mile enterprise features, things like role-based security, Kerberos, LDAP and getting all of those plugins available so people can actually get it in a production environment. Yeah, it's a tough challenge. I mean the VC community, if you overshoot on promises you have a cut to your head with the investors. So it sounds like you're laying it out saying we're taking a pragmatic approach to the code base, get it out there and then develop some value and then figure out where we can add value from a monetization standpoint. Sounds like what your strategy is, right? Yeah. Somebody like that, but that's just a good approach. Yeah, and we find that Google is very receptive. We joined their startup program, we were able to get $100,000 in cloud credit which helps us to experiments and do things for customers to enable them to get on Kubernetes faster but also as people come to us with opportunities around Kubernetes, I typically bring Google into the mix from day one and work with their partnership teams to do that because I want to make sure that they can drive engineering internally at Google to kind of close the gap. It's interesting, being kind of in the industry and being historically 30 years in the business and looking at some of the recent movements like the Cloudera for instance, I mean no one knew what Hadoop was when they got funded. I mean it was really more of a bet on a direction that made sense. Hadoop, storage, you guys run the same boat. I mean, Kubernetes pretty much is validated. You're seeing cross the board validation of the concepts and some of the code, you guys have some, it's a good team. Similar dynamic, right? So as an entrepreneur, you got to balance the commercialization vision with the directional build out of what Kubernetes brings to the table. How do you reconcile that in your mind? Just kind of laid back and say, hey, you know whatever, we'll just find a partner who understands our vision or use their more pressure to get that cash flow in to get the higher end, kind of like get that Docker flywheel. I mean again, Docker was like, you know, no one heard of Docker and all of a sudden they'd get ramped up. So I think the thing for me is everyone is interested in Kubernetes. If you see the hype that we have here at the summit, everyone's talking about it. So I have so many customers coming to us wanting architectural reviews, wanting POCs and my big issue now is I just physically don't have enough people to do them and I can't do them as quickly as they want to do them. So we're looking to companies like Cloud Technology Partners and things like that to help kind of take our playbook and let them run with it. So we make sure that people are getting the same kind of quality of service but we don't want to keep anyone away from Kubernetes or the community. You want to keep the tide rolling in on the growth side so you're enabling people to be successful. Yeah, and that's why the Mirantis unlocked appliance makes a ton of sense to us. Like in a couple months you'll be able to order that one SKU, have Redap come, install all of that hardware and that's your development kit in a box. With two clicks you've got a Kubernetes cluster and it's a really nice gateway into that world. It gives someone access to a sandbox, if you will, instantly without a lot of the patchwork required. You guys are a native into the appliance. Yeah, and one of my big frustrations with doing the POCs at this point, like I'm not going to name names with some Fortune 5 companies out there. They want to do this and they're like, yes we have miscellaneous hardware that we can use, yes we have a lab environment but then we run into, well okay we've tried to provision the hardware, we're waiting on these interconnects. It's going to be three months before we get that done. So like back to the Mirantis type thing we've redapped. Or just order this one SKU, have Redap come, set it up in your lab and we can be out there next week and start doing your one week POC. And so we just tighten that. And what's that conversation go like? They're like, we're in, we're totally into it? Or are they more dogmatic around the existing gear they got? Well, I mean, just to kind of give overall price points, we charge $75,000 for a one week POC around Kubernetes. And we try to pick one workload that someone has, get that containerized, get that working, and show them kind of the best of breed approach. So when they're already paying $75,000 for one week of our time to do a POC, ordering the unlocked appliance, which is going to be Juniper hardware and Dell hardware, a half rack of that, like it's pretty reasonable. There's not a huge markup on top of just the hardware itself. So when you're asking someone to spend another 300, $350,000 on equipment that they then can use for anything, they're like, okay, yeah, that makes sense. We'll do it. And the Redap people are extremely responsive, so. So what's your vision looking forward next five years? Just shoot the arrow forward. Where do you hope to be? How do you see the industry evolving? If you're floating down the river of open stack enablement and all the goodness that's going on with Kubernetes, I'll say it's a great opportunity for computer scientists. I'll say you mentioned the go language. I mean, it's really great, robust, fertile opportunity environment. How do you see that playing out of the next five years? And what do you hope for you guys to be? Well, I really want to be the company that's providing a lot of value to customers and providing documentation and making sure that people do the right things. One of my big frustrations with Docker, you know, at this point, even though they did an amazing job as a company with their documentation, was you got to keep that documentation up to date. And so with Kubernetes, like the one thing I could fault us for is we don't do a really good job of keeping the documentation up to date. And so, you know, I look at Docker and you find someone's old blog post about how to do something. Well, that's not the way to do it anymore. So you have, you know, architects and engineers at companies that are basing their mental model on this on some old blog posts. And the reality is that the ground's shifted underneath them. They're moving train. They're doing it right. It's one of the big challenges with open source in general is the community's working on it. Who wants to document that? Yeah, yeah. And, you know, and that's why I'm excited, frankly, of things like CoreOS and what they're doing with Tectonic. Like Tectonic gives you Kubernetes, you know, you buy one product from them and you've got Kubernetes, you know it can be updated, you can trust that the kernel updates are going to happen. And if you like that model and you're willing to pay for it, like all of that kind of fear goes away, right? And you've got a nice company to call and get support for it. So I like that. It's an illusion. It's a bridge to cross over. I mean, people, customers want a bridge to someone to cross over with and necessarily they want the trust factor. Docentation and support, packaging. It's just evolution. I mean, so we're early days. We're certainly early days. Great opportunity for you guys. And what I would really love to see, like in former life, I was senior technical advisor at Wikipedia. And that's the fourth largest, you know, web property out there. And I would love to see something like Kubernetes being played, you know, with Wikipedia where I don't have to have engineers or ops people SSH-ing to live production machines to make changes or debug things. I would love that people don't even have to think about SSH-ing to a machine. So I log into the Kubernetes cluster, I update the container to a new version from the container registry and now I've launched. Auto updates across all the different code. Yeah, beautiful. That's a great orchestration. Makes a ton of sense. And some less potential, less error error. It's less error-prone manual error with SSH-ing, anything could happen. And some of the work we've done with our early clients, like Zoom, the payment processing people, one of their Kubernetes use cases is really interesting where it's, we basically want to take our whole production stack and put it in a box so our engineer while he's on CalTrain can actually do all the same work. And when he comes back to his desk, he's going to be able to deploy that to the bigger cluster but know that all of the individual microservices work together. And that's where I get excited too, it's just enabling developers not to have to worry because that time gap between I developed some software and I want to get it in production is just too big. It's too big a bridge to cross right now. And you got to get it to accelerate down. Final question before I got to ask you, just your take on microservices, the buzzword we've been seeing around, Cloud Foundry Summit, you've seen the word microservices being kicked around. What does microservices mean? What is a microservice? Is it a subservice of something else? I mean, is it a new class of tooling? Can you explain that for the folks out there that hear that buzzword? It's kind of in the hipster phase right now of me kicked around. But it's pretty relevant we're finding. What's your take on microservices? Well, I think most organizations that I've gone into as a consultant and various other roles I've had in the past, you see these old apps that evolved really slowly and they have all of these hot spots around, maybe they're using Oracle for a database, maybe they've got Memcache for caching and it's all kind of cobbled together and very brittle. And so when I think of microservice, I look away to like, how do we make sure that each individual piece of our infrastructure can scale independently of each other? And how do I decouple those hot spots? And so to me, it's mostly about decoupling. And I think if I were going to start some new project right now, and I was a web tech company, like of sure I'm going to go the microservices route, but beyond even just writing my software that way, I need to make sure that I have a development stack that's going to make sense. And that's where containerization is exciting. You don't end up in an environment where you've got 400 front end Apache web servers that only can be Apache web servers. You run an environment where like, well, we're trying to patch you right now in containers and we're going to try engine X in containers and see which one works better for our workload. I mean, this is the future of software engineering. It's always been the ethos, decouple and make highly cohesive elements. And that's kind of where we're going with that, right? Yes, yeah. All right, Patrick Riley, CEO of Kizmatic on theCUBE, sharing his insight, citing new company, new startup, great background in web scale, large scale, hyperscale, all the scaling and certainly now bringing that to the cloud. Distributed job. Distributed computing, microservices are all, this is like going back a decade, those are the same concepts. Thanks for joining us on theCUBE. We really appreciate your insight. Thanks for having me. We'll be right back after this short break. This is theCUBE live in British Columbia in Vancouver. We'll be right back.