 Live from San Francisco, California, extracting the signal from the noise, it's theCUBE, covering DockerCon 2015, brought to you by SiliconANGLE Media, with special thanks to Docker. Now your hosts, John Furrier and Jeff Frick. Okay, welcome back everyone. We are live in San Francisco for DockerCon. This is SiliconANGLE Media's theCUBE, our flagship program where we go out to the events and extract the simple noise. I'm John Furrier. Co-hosted by Brian Grace Lee. Cloudcast, welcome to co-hosting. And John Willis, Technical Advancement of Docker. Welcome to theCUBE. Yeah, great to be here. So we're like live, but this is like a podcast, live podcast to here. Hi, mom. Can't take it back. Once this goes live, I have many blunders on theCUBE. So I got to get you guys take on the news here. What's the hot news around the Docker containers and the network? We heard someone talking about plumbing, the OCPs out there. You guys talk about plugins, Prydos, us going live. John, tell us what's the new update there? Well, so to me, I mean, my focus has really been on, so I was part of the socket plane acquisition. I was one of the founders of the socket plane. And what we were trying to do is kind of bring native SDN to Docker. That was our whole mission, right? Very much open V-switch, native architecture. So we got acquired by Docker. And so one of the things in our story, at socket plane, it was very much a David and Goliath story. We were going to go after the big guys. Like we were going to go to the biggest enterprises and go say, this is how you do containers at scale. Because as anybody could see, right? Like containers are great and our basic networks work for works. But if you saw the future of where this is going, the density of compute on a host, right? Which is the new edge, right? It wasn't just going to work. And we saw that very clearly. So when we got acquired by Docker, the idea was that, yes, we love what you're going to do. We love your team. Madhu Venkapals was our main architect and we had some ex Cisco people. By the way, Madhu was one of the original committer on Open Daylight, right? And they said, it's great, but what we need to do is Docker is a bigger, bigger story, right? Like we have customers that are going to want network connectivity with ACI and Cisco. We're going to want customers that are going to want network VM, you know, VMware and the NSX, right? And then Weave and all those things. So what the team did really for about, somebody's giving me eagle eyes over there, but what the team did for the last three months, which is pretty amazing what they accomplished if you listened to the announcement yesterday, they basically built what we were trying to do as a David and Goliath story as a general purpose architecture for the future of networking with Docker. And I'm really, really excited about what that team has delivered. One, we talk about standards of people able to work in conjunction with all the complexity of integration. We have this plug-in architecture and that was a core concept that, that we built an abstraction layer and this primary abstraction layer is a new project on GitHub called LibNetwork. And all the network people are very excited about this. And the nice thing is LibNetwork actually built a separation between Docker. Docker was very tied to network componentry. You start a network, a Docker container, it created a network namespace. I mean, it was very tightly coupled and there was things about that that involved the application developer in a world that should never have been involved. What Madhu and team have done is built an abstraction layer in between the Docker container and the drivers. And it's called LibNetwork, the service abstraction. And so now what you have is the service abstraction takes care of all the glue between anybody wanting container. If I want a container and I want to move it from this network to this network, everything happens to the LibNetwork service abstraction. And if the network in production is NSX and it's weave in some other test ground or all that stuff is completely separated from the... Okay, so yesterday we reported, Stu was here. Stu would be, by the way. Yeah, Stu would be. Stu would love you. He'd be banging the table right now. So, but the big three things from yesterday that I took away was open container standard. Okay, the cross network host support, cross host support networking and the plugins. Tease that out, explain those three things and how they relate I kind of got lost, is it networking? Where's the plumbing? So can you break those three things down? Container standard, cross host networking and then the plugins. That's all working in reverse order. So the plugin architecture is something that the Docker has been working very hard at for quite a while now. It started with storage. And then when, again, when Socketplane came in with a network story, it was, okay, we need to use that similar architecture plugin for network vendors. So the whole plugin architecture is, again, very much think of it. There's plugins going on for storage, plugins for other, but in a network space, in general, let me just set back. In general, the plugin architecture is really, we got a batteries included or you plug in this for storage, you plug it in for network. So that's the plugin story. There's a specific network plugin storage, which is what I described earlier with each of the vendors like NSX ACI. So that's plugin. It's very compatible. Anybody can plug and play. The network, which is the second part of that, was this lib network. This lib network introduces an added box, batteries included, multi-host support. That's one of the things you didn't have. This is what our first order problem at Socketplane was to solve the container multi-host networking. And so what lib network has done now is you have your default bridge network, but you can also, out of the box, specify an overlay network. And the overlay network will actually add it a box, batteries included, allow you to create overlay VX lands, and voila, as of yesterday, we have full support for VX land multi-host support. So that's that. I'll be quite honest with you. I've been so focused on the network. I have not been focused on the OCP. I don't, Brian, you might be able to add a little more. I think the way I look at it is, you know, it's been a set a bunch of times. The big theme this week is Docker and production. In order to do that, you know, there are so few green fields. Everything's sort of some shade of greenish-brown. You've got to be able to work with a lot of things, right? So the plugin thing is really important because people will say, I could do some new things, but I also want to work with what I have, right? Nobody wants to be locked in, at least in theory, you don't want to be locked in. That's the whole batteries and pluggable. And then I think the other thing that came out of all the announcements, whether it was yesterday or today, and you're wearing a DevOps shirt, right? DevOps is this great theme or concept. But people don't always know, like, how do I get there? I think Docker laid out this really nice model that says, like, here's the technology. Here's the build, run, ship concept. And here's how you sort of lay all those things out. And you can kind of see the blueprint for how to do that. Yeah, I mean, it was, you know, I think all the gaps, you know, all the gaps, I mean, we're a two-year-old company, but two and a half-year-old company. But a lot of the gaps got filled yesterday because, again, when you thought about networking or you thought about your network-specific solution or you thought about, you know, I mean, from networking or from storage or and even from a delivery standpoint in terms of, you know, what if people want to use other container solutions, right? So now it is really, you know, if you think about Docker is putting almost the canonical Linux chain in place, I can plug and play any piece of it. You know, the whole beauty of the original Linux toolchain model was, right, that every section of it was plug and play. And we kind of have that now for compute. We have our network, you know, we have our storage. So we got some commentary from the crowd chat around the plugins, multi-tenant, restart patching, no need for that. And what's that take on that? There was a comment I saw from someone from Docker and said plugins don't need a patch and are multi-tenant. What does that mean? Well, so I'm not sure. I'd have to leave a little more context about the don't need a patch. I get, oh, well, so I think I'm thinking, I'm assuming what the question means. If you want to use multi-network, multi-host networking support today, it's in 1.8. So our current release is 1.7. It's a 1.8 experimental. So that means you can download. We have a blog, in fact, I'm gonna do some videos on it this morning, this afternoon. But we have a blog article that tells you how to do install from experimental. You'll actually be running an early release 1.8. If you know that our release cadence is every two months, 1.7 was dropped about a week ago. So figure out two months from now, you'll see 1.8 drop. Probably, I'll probably get beat up by somebody, but I would say 1.9 will probably, no commit, unless I'd be a stable release for this. So if you want to start playing it right now. The really powerful thing about this is, and this is the Docker portability piece, right? If you're the production team, you could run Docker 1.6, maybe 1.7, or whatever. If you're the team that wants to go experiment with things, it's all software. If I want to go run it as an experiment on AWS, I can go do that and emulate the environment. And then, and you talk about this, it's immutable. It means it gets consistent. I can then bring that back into my production environment exactly the same. And that's so powerful. I don't think people totally grasp it, but that concept of sort of dev and test and prod is so consistent and run it anywhere. I think, again, focusing on the network side, but I think it's a common theme is, Madhu says that we want to do for networking what Docker did for compute, right? So to your point, that- Make it easy. It really makes it easy. They don't have to actually worry about any of that. They really literally, again, I don't have to sit down and have discussions about, are we going to choose NSX or Cisco for networking? The application people really won't have to worry about that. We say that's true, but if you go to a larger organization and you go sit through a bake-off between a new eyes or a Cisco versus NSX, everybody in the building is in those discussions, right? In this model, the truth is, the delivery model is with the plugin architecture that the application people really don't, like Madhu says, all they worry about is services. Yeah, I totally, this is the new, this is what to your point earlier, this is the new normal and your kick-off comments around your vision of the company before it got acquired was you're going to take on the big guys, big whales, and just drop the hell out of them and innovate. So I got to ask you guys both, this innovation cycle that's upon us, I was speculating on the kick-off, like, this is the beginning of the front wave. So what's that new normal? The numbers are clear. Look at Amazon's numbers, even Oracle's throwing some numbers out. It's a big guy. Their growth is phenomenal. The consumption by the customer is no-brainer. They're not going to buy servers and storage separately. It's all kind of going to be integrated. So this huge ease-of-use theme, code that runs, they don't want to buy separate code bases. And then get cloud now is an opportunity with economics and performance. So performance and economics are the key thing. So what are the key things from your perspective that's going to really throttle the straight and narrow on the innovation? Is it the containers? Is it the network? Is it the management? All of the above? So I'm a DevOps guy. I've been a DevOps guy. I was at the inaugural DevOps days. I was one of the core founders of DevOps days in the US. So I always say that there is always a cultural afoot that has that, right? In order to change your mindset about delivery, that's the biggest problem, right? Change from a monolithic legacy model of, you know, to a cooperative. So that's, so let's get rid of that. But after that, then there is this incredible convergence of containerization, microservices, and data gravity. And I talk about a lot about this. And so what you see that, you know, microservice, so microservices and containerization are just a natural fit, like, and microservices to the application development world is really just a SLA version too. Define microservices from your perspective. People have different definitions. What is microservices? So microservices really, there's a really good book on Riley by Sam Newman called Microservices. But if you had to boil down to two definitions, one, I think Adrian Krakow has probably the best definition. He says it's a loosely coupled, service-oriented, and I won't mention it with bounded context. And the key two points there is loosely coupled and bounded context. Loosely coupled means that you are building, so you're thinking of those two things. I think you can get your definition clearly. Loosely coupled, you're defining an architecture of delivery of software such that I'm creating my own kind of business-related services and you can use Java, can they see me point? Yeah, just call me Bob, someone was calling me Bob. Bob can use- You can use Java, you can use Python, you can use Ruby. Like, we've fully decoupled the monolithic of how we deliver things. If your group finds that this framework is good, that's great. That's one part of it. I think there's, the second part of it is bounded context. Bounded context goes all the way back from what we've learned over the last 10 or 15 years about software development. Domain-driven development. And cohesiveness or impunctionality. So it becomes the, you start running in on this idea of building a bounded context around a business piece of code. I've got a certain business piece that I need to solve. That's the boundary. Can you describe that code as a business service? So if you couple, loosely coupled with bounded context, and then Sam Newman, who is the author of the book says, by definition it's small, right? I think you throw that together, there's a lot more. It's 10 years of thought leadership around development from guys like Martin Fowler and people at ThoughtWorks, Domain-driven, Eric Evans or whatever the Domain-driven development. Heroku's 10 factor- Tim Crawford calls it SOHB2. I think here's the real takeaway. So you look at who is the customer speakers this week, right? So you have a few that are unicorns, but you have Capital One, you have GSA, you have some other. The takeaway is, doesn't matter what business you're in now, anybody who's a line of business manager, somebody with a great idea can go off. There's no friction to doing idea to execution. And I mean, you guys are a great example. You were six, seven, eight people. You're going to go bust up a humongous industry, but you could be in banking. You could be in government. You could be in energy. And the ability for you to go transform or even just disrupt an area, could be just one project. When there's such low friction, whether it's containers and microservices or it's using public cloud, that changes all the economics. It changes the speed of the game. And once that speed starts happening, then you have to reset everything. You have to reset all your thinking. I think we've used the phrase in computing Chevin's Paradox for a while. And I know Bernard Goldman just wrote a article about that. But going back to that, if you take that kind of convergence of microservices containers, there's an insatiable need. There are a bunch of SLA people that are just waiting for a framework or a technology to deliver something that now containers just... It busts out. Basically, it's just like breaks out of the box. Exactly. But the third piece is just as important in this exponential growth model we're going into, which is data gravity, indirectly IOT, right? So what you find is, Brian Kent Shelf of Joyant talks about, like we're going to get to a point where you're not going to be able to move the data. Right? So the whole paradigm of how we do compute, you know, classically our industry, we move data to compute. We even use load bounces and all that. When we get into these like just swarms of analytics and data that we're going to start collecting in IOT, and we are collecting, banks, fraud, internal fraud, all this stuff, it's going to be harder and harder to move data. So now what we're going to do is, and this is Dan McCrory's theory of data gravity, is that what you, in simple terms, you basically start moving compute the data. So you think about your geography of data. This is your density point. Right, exactly. So now think about this convergence of microservice bearing the contract, containerization of very small, we're talking about things that instantiate in 500 milliseconds, right? And now the idea that I might want to run a thousand years in some geographic area around the world and then run a thousand years in some other year and then aggregate almost like a macro. I mean, it totally nailed it. I mean, I agree with Yondra, sent on this edge of the network because what does that mean now? It's everything. The edge of the networks are a multi-fold. And those three things change everything about what we're going to be doing in over the X5. All right, I want to take a philosophical question. Let's step back. Let's kind of smoke the piece pipe and kind of look at the big picture. The big picture is, every time there's been an architectural change in the industry, massive opportunity has been enabled. I mean, real architectural change, not the .com bubble. That was just BS, right? The fundamental web services was developed. No underlying infrastructure changes. Infrastructure shifts, client server, mini computers, let's go PC, networking. Weird cloud is that now, essentially that. So I want to get your take. What is going on right now in your minds as industry participants that is a radical and or changing underlying infrastructure? Is it the fact that the hardware vendors or selling boxes did converge really play out? Is that really an underlying shift? So what's the big picture in your mind? What is the underlying radical infrastructure change? I think we're, so this is going to sound really crazy, but what the hell? I think we're hitting version two of everything that's happened over the last two years. We're hitting version two of cloud. It's called containers. We're hitting version two of SDN. I'd like to think it's live networks. We're hitting version two of SOA. All this is converging right now, right? Like, let's face it, five years ago, the world, six, seven years ago when Amazon put a stake in the ground, the world changed. And we've been in this learning code for like five, six, seven, whatever it was, whatever, 2007. Just say seven years. Seven year itch. We've been in this crazy kind of learning opportunity and I do think, I agree with you. So I was there, I'm an old guy. I actually started out on IBM mainframes working for Exxon. I was assistant programmer at IBM mainframes doing assembly programming at Exxon and I watched that whole distributed compute happen where those mainframes were getting, literally, crowbarred out of the building, you know. And people moving to new skillsets. Mind shift, mind shift. That was a major, the internet bubble was interesting. But it was no underline, no underline, I agree. I think what we're seeing right now is this opportunity for, again, I'll go back to the data gravity microservices containerization. I think this is a fundamental shift in what we're going to look like from an infrastructure perspective. And do you think that companies that don't adopt cloud are going to be completely missing the boat? I do a presentation called, so a presentation I've been doing called Guns, Germs, and Microservices based on Jared Diamond's Guns, Germs, and Steel if you haven't read that book. And he talks about the haves and have nots. I firmly believe that the new Guns, Germs, and Steel is microservices containerization and data gravity. And the footprints are changing. So you say on premise, cloud, yeah. The old telephone closet days go back to showing our age. That could be the data center now. I mean, you don't need to have the big data center. Cloud is here. The economics are unrefutable. And we haven't even talked about what's happened with density of hardware, right? Hardware and telemetry coming from, you know, Intel's putting almost everything on their chip now. Acceleration, I mean, you know, the hardware is just going to, you know, we saw this, don't put me in a podcast here. You're going to get me to speak forever. I mean, you've got, the economics are, the Cube is just live podcasting. The economics are, little guys can get money. Then you've got the numbers of the community that help build that thing. So they don't have to own the developers. They don't have to be a huge company. And they don't care what the old guys' rules are. They're going to go drive forward. They actually are sort of building the new rules and all the large, older companies are going like, how do I play? And so you get these, like, there's no friction in theirs and there's tons of friction on the other side. And it's, I mean, big games are competing against the hardware. I worked for IBM right at school and then HP in the late 80s, early 90s. And, you know, we were in HP shop. We were in IBM shop. There's no shop anymore in my mind. So what I'm worried about is all these hardware guys, there's, I mean, talk about architectural change, you know. If you're not aware of this. So the question for you guys as thought leaders is this, where's the hard and top? And I had this argument, no, not argument, but a conversation last night with Dave Vellante about this. Do you care that it's an Intel or AMD processor? Yeah, at some point, Intel was a good processor and it worked, but there's a lot of proprietary code. It's a hard and top. Is cloud going to have a hard and top? And where is that? Cause the commodity's happening, I agree, but it's not a race to zero. The, the value will shift to the apps. So the question is, where's that hard and top? Yeah, I mean, I think abstraction is already the value, right? I mean, to your point about, you know, what you can do now, how cheap, right? So I'll give you a, as you said that, I just thought of an interesting meta point, right? So there was a company about five years ago called FlightCaster. They were a couple of ex-quants, right? Like Wall Street quantitative analysis guys, right? And they just wanted to get out of the business, right? And they got it, they did a, this like funded, what do you call them, the accelerator. They were in one of these accelerators and they built this thing called FlightCaster. They did flight predictions and they did it all on, they did it on Amazon. They used to dupe. They used a whole bunch of Erlang. They used like really cool technologies, Co-op, closure, everything. And I got them, I used to, I used to do a clock, if they parked because I don't do it anymore. And I got them on the podcast and I said, how much did this cost? And it was like $180, $200,000, right? And the thing was, and it took them like four months to do it, right? And I used to go out to conference presentations and I'd say, this is the world looks and I'd say, imagine if Delta and American Airlines and all got together in collaboration to do this. I'd say it would take them three years and it would probably take them 10 million. Guy comes up to me, asked one of the presentations, you were wrong, because I worked for American Airlines. I was like, oh. It'd be 20 million. Yeah, exactly. It'd be 20 million. I was a judge for the Dakar Hackathon this weekend. There was this one project, it was about three or four projects, the three winning projects. One of them just blew me away. So, and they came in like third. And I asked them, just like, it made me think out of curiosity, how much did that cost you guys? So they built in 30 hours. They went ahead and it cost them $80 in Amazon fees. And they built a product that literally probably could get funded tomorrow and has an eight value. Now, why does it have that? Because it was on the shoulder of giants, right? They did, again, back to the Jeffers Paradox. There's got a supercomputer in the cloud. They had Amazon, they had Docker containers. They had Docker files. They were using all the Docker latest tools. They were conductors of code. I mean, they were essentially spinning up resources like it's nobody's business. But that's the world we're in now. Totally agree. I totally agree. All right guys, we have to wrap there, but I want to get your final thoughts, both of you guys. And I want you to share with the audience, the folks who are not here on site in San Francisco, who might be watching remotely. What is this show all about this week? What's the big takeaways? You could share some color around the important things going on here to show the vibe, just your point of view. It's Docker, Docker, Docker, right? That's the track I run, right? I've been memeing the Docker. But the thing is that Docker's for real. The way Solomon and Ben and the team are thinking about how to do this is real. The government is playing. I love when Scott Johnson said, I know Scott from Puppet and many of his startups before and he got up and said, our first commercial customer is GSA, right? So my takeaway here is Docker is for real. It's not like, if you're sitting outside, oh my God, all the tweets from Docker. I mean, yes, you know, we're pretty excited about it. Yeah, it's intoxicating. But it's real. I mean, what we're doing here and what we're doing technically, what we're presenting, we're going to have one in the EU, we'll have one obviously next year. I see, you know, get your butt over to DockerCon because it's, I don't think there's a, there's a- It's a small community is growing very fast. They're in an unhappy face in this hallway. Brian, your thoughts? To me, it's two things. There's a massive blending of devs and ops. So I mean, you walk around and it is blended together. It's not a segmented show. And there are so many people excited about, like I can go build apps to do, drive something for my business. And I mean, like just that excitement of going like, I'm going to go affect the top line of my business is amazing. And obviously the numbers of the show are packed. It's crazy. I love to go for the business intelligence guy in the keynote this morning. He said, he said, why we chose Docker. It was fun. Yeah. And it worked. And it worked. And it worked. Yeah, that was the second more important. But the fact that he said that it was fun. I'm like, yes. Yeah. Stu Miniman from Wikibon says, it's the center of the universe. It's really app focused. All the energies around the good things in life, right? Fun, cross portability, compatibility, no lock in. Creative ideas, all those stuff. So this is the Docker. It's free love. I call it the Woodstock of developers. It's really a, it's a cool environment. People are cool. Things are getting done. And it's a ton of opportunity. It's at the beginning. So we'll be more, more coverage from theCUBE after this short break. Stay with us. We'll be right back after this short break.