 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, Stu Miniman and Jeff Frick. Welcome back to SiliconANGLE TVs. Live, wall-to-wall coverage from DockerCon 2015. Here in San Francisco, I'm Stu Miniman with wikibon.com. My guest host for this segment is Brian Gracey from the Cloudcast. Brian, thanks for joining us. Thanks for having me. All right, and we've got on our segment and what I wanted to bring Brian is we've got Adrian Cockruff, who's a technology fellow at Battery Ventures. You know, so many of these new technologies over the last couple of years. They talk about what's happening in cloud, containers, and microservices. You know, Adrian's sessions are some of the ones that help educate me, help educate the community. So Adrian, I know you're swamped at this show. Appreciate you taking time to join us again on theCUBE. You know, welcome back here at Cubelon. All right, so tell us Adrian. You know, what excites you about this show? You know, you've been watching this space since, you know, the early days. Heck, we were talking back, you know, about Sundays and how, you know, Solaris containers were around, you know, long before Linux containers were. So what's your take of the show so far? Well, what I think is particularly interesting with the Docker ecosystem is the speed it's moving at. And that's really unprecedented. It's moved, typically these kinds of technologies take years to get from an idea to be something that enterprises have gone for. And, you know, one of the interesting things that I saw this morning in Ben's keynote was, I think 97% of enterprises are adopting or want to adopt Docker at this point and wave way earlier and much higher penetration than anything else in the history of kind of new technologies coming along. So that's what's particularly interesting. Means that the whole ecosystem is changing week by week. There are new announcements every week. There's a lot of announcements this week, but really every week there is something large that changes the ecosystem. And it's fun for me to just keep surfing this wave that keeps changing and you have to be very agile if you're going to operate in this space. So that makes it fun, right? As a VC, it means that everything you know is wrong once a week means that you have lots of opportunities to try and find, okay, who is actually going to break through and be the next interesting piece of this ecosystem. And that's my new job. Sorry, Adrian, last year at the DockerCon show in San Francisco, Docker 1.0 released. We're up to Docker Engine 1.7 and we've got all these other pieces from machines, warm compose and all the others. So where are we with the technology? How robust is the stack? Where are the big gaps and where are some of those major opportunities that you see today? Well, I just gave my talk and the summary of my talk was that in 2014, there were people running Docker in production and it was really DIY frameworks. In fact, there were people in 2013 running Docker in production. I think it was Gilt and Drama Fever and a few people very early on went, okay, that solves the problem. I was trying to build something to put, to build a framework anyway. I can use Docker as a piece of that framework to go and put something up. 2014, a lot more people copied them and what you're seeing this year is that the systems are starting to mature. So the best practices, the security white paper, we're getting standardization, really big announcements today, things like having pluggable ability so you can now run cluster HQ and we've at the same time on Docker instead of having one or the other. So those kinds of things are really, it's starting to mature to the point where you can build standard practices and tooling around everything. And I think during this second half of this year, you'll start to see many more big enterprises start to adopt and really roll things out. And in 2016, it's going to be pretty mature, very well standardized, the good practices are in there, people know what they shouldn't be doing and it's getting there. And that's pretty rapid move for software and the enterprise space. Yeah, so you've got a slide that you use in some of your talks that said 2014, nobody had it in their roadmap, 2015 everybody's got it on their roadmap. They're releasing stuff every two months. If you put that in perspective, VMware is once every 12 months, OpenStacks every six months, they're every two months. How do you, and we've got all these sort of kind of overlapping projects. You've got multiple network ways, multiple service discoveries. Do you think as this becomes more mainstream, people will just sort of standardize on a stack or do you think we'll keep having the developer mindset of lots of tools or how do you start to see that ecosystem shake out? I think it becomes all things to all people. It's not going to be one stack. It's not going to be one architecture. There are people that are going to write their coding go, do a static build, build a scratch container that contains just one binary and have that be their container. And it's tiny and it's fast and it's secure and you can't, it doesn't have a shell. You can't SH, there's nothing there to log into. So that's at one extreme. At the other extreme, somebody's going to look at this big pile of crud in their data center and says, you know what, I could run all of that in one container. And I got this, you know, I maybe got an enterprise service bus lurking in the middle of a container or something insane like that. But you could wrap everything in it. But I get value from the fact I've now wrapped this complicated configuration into a container. I can now move it around. I need this thing. Maybe it's my point of sale thing I have in a shop and it contains pile of old junk, right? But I have 600 retail stores. I could now deliver that in a containerized way. There's people that are going to be putting all kinds of stuff in there. And then you've got the VMware world really starting to adopt containers and VMware making it easy to have containers be first class citizens. You've got all of the sort of past vendors adopting it. And so I think it's going to be everything. It's not going to be one specific place. But the thing it enables that's going to allow new architectures is these extremely rapid response systems. Because you can fire up containers in seconds, you can start auto-scaling in seconds. And I think if you look at AWS Lambda, which is not a Docker-based thing, but it's a container that gets fired up to run one request and then shuts down again, that's sort of the other extreme, right? You now have containers that are completely ephemeral. So there will be possibly systems getting closer to sort of the Lambda-like environment starting to come up and those patterns that look like that. Yeah, yeah, makes sense, makes sense. Adrian, can you tie in, where do microservices fit into this whole discussion? You said I can put kind of everything in a container or I might have an application that's spanning between multiple containers, which boy, if I think about how my infrastructure works, that spawns a whole separate set of discussions, not the security, but if I have any data that I need to maintain, how does all that play out? Yeah, so microservices is really, it's a tactic for achieving a goal. And the goal is speeding up development with large teams. If you have a small team, you build a service. They can talk to each other, they can figure out who broke what. As your team gets bigger, you start to have developers and operations and testing all starting to collide and it gets harder and harder to get a build done. And you need more and more time to test because you're trying to integrate all these things together. That's the point at which you break things into microservices. And microservices, they're less efficient. They use more memory, they use more CPU, you've got more things going on, they use more network, they're a little bit slower, but you're paying in the efficiency, you're trading off efficiency and cost for the ability to move faster. And the problem that most big companies have right now is they're moving too slowly. The problem isn't that they're spending a little bit more money on a particular service. The problem is that they are releasing a thing once a quarter and their competitors doing it 10 times a day. So in order to get to continuous delivery, you have to break it into chunks that are manageable. And microservices comes in as a way of separating the concerns so you can do that. So you obviously had a bunch of experience doing this at Netflix, it's well documented. As people talk to you, they come up to you after a talk or you're out talking to clients and so forth, you guys went through a process of kind of matching the Netflix culture with a business need. When people come up to you and they say, look, you've got some experience doing this, like where should I start? Do you ever have a starting point that you point people to? Is it one service? Is it does certain company cultures fit away of starting a service? What's a, how do people get started down that path? Yeah, there's normally the best thing to do is there's a couple of good books, right? So if the whole discussion sounds like there's a bunch of aliens from another planet wandering around, then you really just start with a Phoenix project, right? Start with a novel about a company that is just can't figure out how to build software and it's going to go out of business or it's going to just lay off its entire IT department and it's a horror story, contains all the bad things that anyone has ever seen in production. So that book really sets out, okay, if this looks like your company, then here's some patterns and ideas and at least some hope that there's a way out. If you're a bit further down the road and you've got, you're just trying to look at how do these parts all fit together, then I think the Lean Enterprise book is the key one and I've been pushing that book for a long time. That's a hypothesis driven development, how to organize things. There's a lot of detail in that book on how enterprises are really going and moving into this new world, DevOps, Agile, and continuous delivery as ways that they work. And if you're already into that and you're trying to figure out, okay, so how does the microservices piece work, then Sam Newman's work on microservices is really detailed stuff, like how do you break a database into lots of little pieces? How do you think about building those bounded contexts to separate stuff? How do you reorganize teams? So that's much more hands-on. So those are kind of the three books that lead you, I think, through the onboarding process. Makes sense. So Adrian, your current role, you're a VC. One of the big questions a lot of us have is, how's anybody going to make money on this stuff? If you're the center of an ecosystem that's growing at a ridiculous rate, you can kind of leave the making money to a little bit later. So I think Docker themselves are, you know, there's just a massive central ecosystem in that. So the interesting thing now is, how do the little companies around the edge of that ecosystem play with Docker? And it's tricky, right? So you've got to find a gap. You've got to figure out how to monetize it or maybe you're going to get, you know, as the ecosystem shifts, you could be flavor of the month, one month and be nothing the next month, right? Or, you know, Docker adds a feature. So it's a difficult and interesting area to navigate right now in its very formative stages. But, you know, the growth rates that Ben put up on his slides were just, you know, I tweeted that it went beyond ludicrous as plaid, I think. 18,000% growth in Docker, yeah. Okay, yeah, that's not a usual percentage to put on a graph, right? It's a ridiculously high growth rate. Once you've got those kind of growth rates, you know, it's more about trying to make sure you don't mess up the ecosystem. And I think the RunC getting the CoroS guys back on board, making sure that they don't split the ecosystem. They have everybody bought into it. I think they're making the right moves to keep this as a single ecosystem rather than having it fracture, which was kind of the worry six months ago around DockerCon Europe. So that's, I think that's a great sign of kind of the maturing that little ecosystem into something where everyone's bought into it and people's concerns are all being heard. They've got the foundation. So that is, I think, a really fundamental announcement. Monetizing it once you've got all this stuff set up is something that's going to come to them later. I mean, it took VMware a long time, I guess, to become a real money making machine, right? Absolutely, and VMware got acquired before they were making tons of money. So one of the questions I have, you're working with all these startups, the big companies are already here. If you look in the cloud space, I mean, IBM with BlueMix, AWS and Azure are here. Big storage companies here. The networking guys are all partnering with there. So is there room for kind of a small growth of startups to actually monetize and make some money? I mean, the open source, they can obviously contribute and add value. And we see lots of projects out there, but you know. Yeah, I think there's two sides to this. One is that the ecosystem is evolving so fast that once a big company gets involved, they can't move that fast, right? So you have the inherent advantage that if you're small and you can go quickly, then you can out execute a large company. If you get it right, and then there's just so many companies swarming it that some of them will get it right. So that's one side. The other thing is that we're seeing rapid adoption by enterprises, right? You have Capital One listed up there as a bank, right? There's a bank listed as a primary adopter of Docker technology. And I've spent some time with the Capital One folks and they are being very aggressive, right? So banks, there's a lot of money in banking, right? They have, when they start doing something, it's a big scale, it's big budget, they get support and they go and buy real technology to solve real problems. So I think that once you see those, what are traditionally the later adopter, high spending businesses moving to this kind of technology, yeah, there is money to be had if you have, if you solve a real problem. But there's going to be, the ecosystem's going to spread out. There's going to be more and more things around that ecosystem. So there's like security companies. How do you build security tooling and auditing and things like that around the Docker ecosystem? Just starting to see some pieces in that space. And there are some vendors here or some startups here that are trying to understand what's happening. I mean, you were talking about security today with microservices, Docker published best practices this week and VMware was making it kind of a big deal about security. I mean, is that the last big frontier technical hurdle for Docker to have mass adoption or are they already there and this is just the enterprise last hurdle? I don't know what the last, whether you'd say it's the last hurdle. It's evolving and each step, it gets bigger. So the adoption gets larger. People get more comfortable with it. People go, okay, I trust VM security. Five, 10 years ago, people didn't trust VM security. So okay, so they got to the point now where VMs are, they understand that it's secure. And then Docker's container security, they're not sure yet, right? After the talk, Brian Cantrell came up and he from Joints said, no, we have secure containers running on Joints. Okay, but most people don't trust container security at this point, but it's probably better than they think and it's going to evolve until it really is good enough and people will be happy with it. So I think that that's something where part of it is just, this is moving so fast that you got to socialize the ideas to a point where people kind of have good, enough good experiences and the patterns look good that people go, okay, I believe this now, right? So the technology is what the pieces have to be in place and then people have to get used to it and then it becomes receive wisdom and everyone gets comfortable and you move on to worrying about the next blocker. What's the next thing that you can't do? Yeah, I mean, one of the customers that they highlighted was the GSA. I mean, if the federal government and the GSA are going to take 10 massive applications and dockerize them, they're not going to do that if there's no security. So the discussion I had with Docker is they said there's the Docker trust model and it's really layers. So does that resonate with your thinking? Yeah, sure, and people weren't trusting public cloud and people weren't trusting and people that won't trust public cloud have like half their data in Salesforce are going, yeah, it's really, and nowadays they've got over that. So people are finally realizing that you can build secure applications in public cloud. And I think there was an interview I did recently where they said, what's the thing in the future that other people don't believe yet that might happen or one of those questions? But what I came up with was at some point in the future, the most secure way to build an application will be containerized on public cloud using very fine grain. You've got identity access management, you've got roles, you've got everything locked down. The only thing this piece of code can do is make this API call to here. That's all it's allowed. This is the way AWS Lambda works like this. So you've got an incredibly locked down application with an incredibly tightly integrated audit capability. That sounds like an amazingly secure way to build an app and if at some point that becomes the only way you can get certified to hold really secure data is to do that. All of that data that's currently sitting in data centers becomes the insecure because it's so easy to breach and get in there and rummage around and it's not really under any kind of detailed control and the auditing isn't really that detailed. It's not audited by design, it's audited by somebody filing a ticket. So once you get to the point where every API call that modifies your infrastructure is fully logged and audited and there is no other way to modify the infrastructure than making an API call. This is potential future world where the most secure way architectures are cloud containerized systems and audit and compliance people would be pushing people to do that because your data center has been breached another 10 times this month, right? To get out of that data center. But that's, I'd say that's controversial in some places but other people are going, yep, it's happening but it's, yeah, that's going to be a while before people get comfortable with that. Yeah, it's interesting, in our survey work talking about people about cloud if I haven't done a major initiative in the public cloud security is the reason I don't want to go there but once I've done it I realize that, you know it might be different than what I've done but it's probably better than what I was doing in house. There's two sides to that. One is security, the most jurisdiction, right? There's the ability to build something secure. I think it's undoubtedly you can do that in cloud. It's very clear. Jurisdictions are doing it. What laws apply to this thing I've built? And if you live, you know, the Germans for a long time are going, oh, we don't trust this public cloud thing. We'd have to put our machines in Dublin, but not sure. And then, you know, AWS opened Frankfurt and I actually visited Germany before and after that. Yeah, we're running in Frankfurt now. They turned up on our doorstep. Okay, this is a thing now, right? And the French are going, and the UK are going where's our clouds? Well, okay, there are some clouds there but you know, the big public clouds are gradually spreading their foot print out. And as your credit have done a very strong thing. I think that's driven largely by need to run things like Office 365 where they want those emails to stay in a certain country. So they're driven by their application portfolio. But I think that is the end game that you're going to see public clouds, small public clouds in every major market because of those jurisdiction issues. There are laws that say, you know, I know there's UK laws that says if you're spending government money you have to spend it in the UK. You couldn't spend it on something that happened somewhere else because that's a misuse of public funds, right? It's not to do with security as such but I think that though as this is all maturing this these things are starting to happen. So we all sit around sometimes as media watch how fast this changes and head spin. Vendors are trying to set strategy. As a VC, when you've got companies that are spinning up as quick as they are like how does your world change when these things go so quickly? Like how much are they leaning on you to help them with strategy? Or, you know, like what's a VC world change when everything gets faster? The fun part for me, I always wanted to know what was going to happen next, right? I originally, one of my first jobs I was a software engineer and we were buying sun machines and I wonder what sun machine was coming next and they wouldn't tell me. So I joined Sun partly so I could figure out what they were building next because you get about two or three years into the future of they were building this hardware that I got to find out about in advance. And one of the things I like about the VC world is I get people that haven't even left to start, you know, they're at some big company and they're thinking about forming a company and they've got ideas. Some of the most fun things we get is to get together with a group of people that haven't even got a name for their company that got a vague idea they want to build something. And, you know, every week or two we get together with that kind of thing and we're brainstorming a possible future company that, you know, if we do it right and we get the right connections in there then, you know, we get to fund this thing and it turns into an actual thing and, you know, helping them through that whole process is very rewarding. The early stage, seed stage, that's, you know, that's kind of the most fun part for me. Yeah. I'm working, we have a new general partner I'm working with, Darmesh Thacker who came from Intel Capital and has done a lot of seed stage startups and early investor in Mirantis and places like that. So he is bringing a lot of real deep experience to the team that I'm working in now. We've only had him since February but we're working through some early stage deals and that's kind of my side of it. So Adrian, we're running out of time. Last question I have for you. If we've gone to plaid with how fast we're moving and you've got this huge ecosystem that's building, you know, can this ecosystem keep running so fast? You know, you kind of say as things go almost vertical, you know, the ship starts to break apart there's things fragment, you know. So a year from now, are we, you know, just an order of magnitude more or there's some kinks in the road? We're going to hit a road bump. You know, what do you think happens in the next year? I think that the, I think we kind of hit the road bump six months ago with the whole, you know, CoroS rocket, APSI, split off where when it, when Docker first happened it was from a non-threatening source. This tiny company I would heard from built this cool thing. Everyone went, oh great, that solves a problem. Let's adopt it. Six months later they went, oh crap, looks like we gave away some crown jewels. How can we kind of get back a little bit of control here? And that's over the last six months, that's now worked out. So now there's a Linux foundation based entity. They've given the code into it. They're everyone signed into it. So they have to execute on that. I mean, they had, I think it's set up right. They've got the rough consensus working code. They figured out the right level of what should be in the foundation, what should be on top. So I think that they've kind of passed that road bump. They've now have just execute against what they've got set out. So this is really happening way faster than most similar kind of environment. Yeah, I'm tired of trying to talk. All right, well, Brian, I really appreciate you helping me out on this segment. Adrian, thanks so much for taking time out of your busy time at the show. We'll be right back with lots more coverage here from DockerCon right after that, this quick break. Thanks very much.