 from Austin, Texas. It's theCUBE, covering DockerCon 2017. Brought to you by Docker and support from its ecosystem partners. Welcome back. Hi, I'm Stu Miniman joined with Jim Copilis. You're watching theCUBE's SiliconANGLE Media's production of DockerCon 2017. We're the worldwide leader in live enterprise tech coverage and we can't finish any DockerCon without having Jerry chain on. So Jerry, partner with Greylock, always a pleasure to interview you. We've had you on the Amazon shows, lot Docker, other ecosystem shows. So great to see you. Stu, Jim, hey, thanks for having me as always. It's great to be here. All right, so first of all, I mean, you invested back in the dot cloud days. Could you imagine when you were meeting with Solomon and those guys and everything, that we'd be here with 5,500 people as to where they go. What's your take on the growth? Every year just blows my mind. Both in open source community, developers, ecosystem partners, and more recently the past year and a half, the enterprise customers that take Docker seriously and replatform the applications on Docker amazes me. I think I did the investment in 2013 and there's a few hundred thousand downloads of Docker. Now there's billions and billions of containers being pulled. When I talked to CIOs, I do frequently through like Docker, containers, what is this thing, pants? And then three and a half, four years later, I can't have a conversation without a 4 to 500 CIO without talking about their Docker container strategy. By the way, I hear if you do send back a belt or something that's broken to the Docker people, they'll fix it for you and maybe send you some whale stickers. It's like the old school Nordstroms where they'll take any return, like this urban store, the four tires will return Nordstroms, return some pants to be fine. We work on container strategy, but we're also your repair shop for men's apparel. So it's always interesting to look at. Integration fabric is what they work at. Yeah, exactly, it's good to date. Brilliant. The maturation of technology, of ecosystem, of monetization, I feel like you talked about the growth of the containers. We've seen the ecosystem has gone through some fits and spurts and changes over the last couple of years. I think we really well received this week and then there's kind of the money maturation and how they mature that. What do you see? How does open source fit into your investment strategy and any commentary on Docker and beyond? Yeah, it's interesting. We're thinking about this on the flight over here today. Open source today is very different than open source five years ago, 10 years ago, it was 15. So what Red Hat did 20 years ago, very different than what Zen tried to do 10 years ago when I was at VMware, very different than what Docker's doing today. And it's different in a couple of ways. I think the way you monetize is different because you have cloud and cloud changes things. The ecosystem's very different because all of a sudden the developers and contributors are not just kind of your misfits and rebels working on the weekends. There are Fortune 100, Fortune 50 companies, their jobs are how dedicated they do this. And then the business models of the developers, each system, how you work with them is very different. So before you had maybe one or two models to make money at open source in one or two ways to develop a community. And we did that at Red Hat with Graylock was lucky to have the investors in years ago, it was at VMware, our cloud founder when we built that, we had a model in mind and we had spring source as well. And I think we're seeing with Docker the past three or four years, if they're really pioneering a way to kind of bring open source and community ecosystem into the next 10, 20 years. So I think it's one to watch. I think some is probably as good as anybody understanding what developers need. Yeah, so a little broader, what's your thoughts on developers today? You actually made the comment coming over. There's two big developer shows this week. You've got F8 and you've got Docker gone. Two very different communities. Right, it's kind of funny, it's kind of the census, do you consider yourself a developer? So if I write a line of JavaScript for my developer, my key sense is yes, right? If like developer is from JavaScript to Swift to Docker to kernel hacking, it's all great. But if you look at those two conferences, you have F8 going on right now and the announcers there around augmented reality and messaging and they try and be a platform, but they're doing many the same things. They have a distribution platform, be it Messenger or Facebook and they're open sourcing technologies around the camera, the lens, the filters to have developers A, go through the channel, B, add apps or widgets and it's really beyond my belly copper head of these filters. But Docker today announced a couple great projects, Moby and Linux kit, much in the same way as trying to give tools to the ecosystem developers to build what they want. I think what you've learned is if you give developers the building blocks or legos as they call it today, they're going to build some awesome structures. Yeah, I mean, Jim, we've talked about coming in here, it's like the role of how data science fits into the developers and there's developers such a broad term as to what we have here. One of the core things I have is that the data scientist is the nucleus of the next generation developer because much of the IP that's being built into applications now is statistical models, machine learning and so forth, driving recommendation engines. But much of that development is being containerized, using new tool kits and so forth, but it needs to be more containerized so you can deploy statistical predictive models, machine learning, deep learning throughout a distributed ecosystem of hybrid cloud to perform various functions. You have to be, right now there's, in most companies, there's a data engineer, there's a data scientist and the two typically work hand in hand. One man does a Hadoop cluster and the other one does the modeling. Does the modeling, so one speaks in R and Python and works in a Jupyter notebook, the other person runs on Hadoop or Database or Redis and the two need to work together and so what you're seeing now, and obviously we're investors in Cloudera, that's another great office source company, what you're seeing now is either A, a set of tools and technologies to either blend the two together in some cases, either enable engineers to be more like data scientists or enable data scientists to be more engineers or also see a bunch of technology tools that say, no, two different roles, I'm going to create tools purpose built for the data scientists and create tools purpose built for the power of data engineer and I think there's space for both, right? So to the extent that you have applications running from like news feed or ads to like predicting my self-driving cars to make a left turn, you're going to need tools that kind of are used by both types of populations. Well, Cloudera I think has now a collaboration environment because of data science. IBM has something very similar with what they're doing so it's a team that has specialties such as coders, such as data modelers and data engineers, point well taken. Cloudera has made a major entrance into that space of collaborative development of these rich stacks of IP essentially that include deterministic program code but also probabilistic models in a deepening stack, yeah. So I think you've seen Cloudera definitely follow that path from Hadoop and low level file system like ACFS to these high level tools for your data scientists that's becoming a platform for machine learning for these next generation applications. And I think you see Docker in the infrastructure analogy doing low level tools like Project Moby and Linux Kit to like high level services around Docker data center, right? And so you can either have the basic tools for your low level developer or for the system admin or administrator who wants to operate and run the cloud, you have tools for him or her too. Yeah. It's interesting, you look at some of these projects and some of the maturation in PivotGC, we talked in DotCloud, went over to Docker. I see a bunch of open stack companies that are now Kubernetes companies. I see companies that were big data, oh, I'm an AI or ML company. So it's always like, is it, it's usually not the tool, it's the wave. It's, what's the driver? Is data the driver of our next wave there? Is it the application? Is it some combination of the two? Those are the two that I usually look at, follow the data, follow the application. Yeah, I would say it's data driving, it's really data application making you, it's data in the application making use of the data. Algorithms, I think, is a component, right? They're important, but they're a component. So what you're seeing now is to be on the right side of history, data's outstripping compute and storage, right? So the amount of videos and central data that we're generating from our phones or cars or homes, that data's outstripping most of the other charts and compute, networking, whatever. So that's definitely kind of a rising tide or a wave, as Stu was saying. Now, how do we extract data or value from this data? And historically, because you didn't have infrastructure in that cloud or compute capacity to kind of make use of this data, it was kind of stranded. And so what you've seen in a generation of technologies like Hadoop or Big Data or cloud technologies, like what Docker can do to distribute your application across a cloud, that's actually enabling you to now build applications to get value out of this data. And that value can be something like forecasting your sales this quarter. It could be about figure which shade of brown belt you should wear with your pants, right? Going back to your clothing analogy. But it could be like, hey, let me build a model around how this car, this drone should drive or fly itself. And so you combine kind of a vast amount of data in your nearly kind of infinite resource of compute with these kind of machine learning or AI techniques and machine learning is one AI technique, but all these other AI techniques, you can build another generation of applications, this new kind of intelligent application to power everything from your home, your car, your watch, or your enterprise app. That's wonderful. Much of the seed change is that less and less actual coding or programming is being done. It needs to be done because more of the application logic is being distilled directly from the data in the form of machine learning. There's automated machine learning tools that are coming up. Google has been a major investor as it stays with automated machine learning. I would say application logic from the insights, right? So in my mind, application logic, application is reflecting business process, right? Hire to fire or to cash. You still need to program the business logic. Data in itself or AI in itself without that context, without that business process is meaningless, right? Just having a model around, you know, a gym or a stew and it doesn't matter unless you're trying to buy something. So Google pioneered kind of machine learning in a workflow perfectly. You're searching for something. They knew who you were based upon in history and they search through the right ads. They say, oh, you really want to buy a car. You want to buy a house. So in the workflow or in the application logic of a search, they use ML to serve you kind of time information. So if now you're in an enterprise, you're looking at help desk tickets, be it ITSM like ServiceNow or support tickets like Zendesk supporting B2C support tickets. That's a workflow. There's application logic to take information on a user or a grumpy customer and then do things like all Mac respond to a help ticket, reset your password, provision a server. And so I think when you have AI or have applications using this data in the context of the business process, that's magic, right? And I think what we're seeing is some core technologies like TensorFlow out there that they're super compelling but we're seeing a generation of developers and founders take that technology, apply to a problem, could be like I said, HR, CRM, ITSM or to a vertical construction, finance, healthcare. Streaming media analytics is a core area where that's coming in. There's a ton of data. Just understand what you're watching, what you want to see. And so you apply things to vertical like healthcare or apply technology to a problem space like media analytics and you have kind of a wonderful application and hopefully a great company. Right. Jerry, we've talked a lot at the cloud shows about how do the startups maintain relevant and get involved when there's all of these platforms. We've talked about Google cut ties. Amazon of course is eating the entire world and everything. Microsoft is making a lot of moves here. How do companies, what do you look for? Has your investment strategy changed at all in the last couple of years? It is, it's don't, think about this a lot in terms of business models and defensibility. And the question goes, what are the sustainable modes you can build around your business as a startup anymore? You see, feel like a kind of scale and ecosystems, network effects, those were historically big defensive modes for like Windows and operating system. Or now those apply to like Facebook's platform, Apple's platform or AWS. They have scale and they have network effects for the ecosystem. So now your startup is saying, okay, how can either I A, overcome those modes? Or B, how can I develop my own IP or my own motor around myself that I can actually sustain and thrive in this generation? I think it's got to be, you got to play a different game. Like as a startup, you're not going to try to outscale Google or Microsoft, right? Leave that to Amazon and there's three or four players. But you can get scale within a domain. So either a problem space like autonomous vehicles, security is a great one. Or a vertical like construction or healthcare. And you can find, you redefine the market that you can kind of dominate, you can build your own motor around that IP. Yeah, it's interesting. Did you hear Adrian Cockroft who, you know, went from the VC's on the IP and now over at AWS. And he's like, well, rather than go start up that business, come build that next thing at Amazon. And we'll do it there. Is that, is that a viable way for, you know, people with the entrepreneurial spirit to go, you know, be part of that two pizza team doing something cool inside a large platform? Oh, I think Adrian, probably his motivation to get more developers on Amazon now. But I was saying like most of our companies not all by a lot of them start on Amazon. Some start on Azure, some start on Google. Some start with their own data centers. I think what they believe is they'll get started on these clouds, but I don't believe this. You know, we talked about this before. It's not a one cloud rules all world, right? I think there'll be three or four, if not more clouds and every different geography from Europe to Asia to Russia to US will have different clouds, different players. So I think it's fine to get started on Amazon and be a two pizza team with your two pizza team. But over time, you know, I see these applications being cross cloud. And that's where some like Docker comes into play, right? Docker let's it be cross cloud, better than any other technology out there. Yeah. You know, on some level, I mean, actually the mode could be, well, increasingly is the training data that drives the refinement of your AI like Tesla is a perfect example that the self driving capabilities that they built into the vehicle, they have now a few years worth of rich test data, training data, I should say, that is a core mode in terms of continuing refinement of those algorithms. So that gives you a sort of an example of some startup might come along with some very specialized application that just takes the consumer world by storm and then they build up some deep well of training data in some very specialized areas that becomes their core asset that their next competitor down the pipe doesn't have. It has to be a set of data that's unique or proprietary. You're not going to basically out train your model on cat photos than Google, right? So it has to be a combination either proprietary data or a combination of data sources that you can stick together. So it's not just one data source. I believe you have to combine multiple data sources together. Yes. So, Jerry, sitting over Jim's shoulder is VMware's booth. Oh yeah. I haven't talked about VMware at all this week. You worked at VMware. I've worked with VMware since pretty early days. What advice would you give VMware in kind of the containerized cloud future? How should they be doing more to be part of more conversations? Well, I think it's amazing that they have a presence here in the size of scale because the past couple of years they've really done a lot to embrace containers in Docker. So I think that's first and foremost. They've done a couple of great moves lately. Embracing Amazon last year with VMware on Amazon was a big move. Abracing containers with some of their cloud native technologies I think is an aggressive move too. So I think they're moving in the right direction, right? So I think what they need to understand is are they going to revolutionize themselves? Are they going to push these new technologies aggressively? Or are they going to keep hanging on to some of their old businesses? And for any company of their size and scale, they have multiple motivations, but I think they're making the right steps, right? So five years ago or four years ago, I don't think they would have taken this DockerCon seriously. I don't think they were exibbers of the first DockerCon, right? But the past call it 24 months, they've done some amazing moves. So I would say it makes me smile to see them take these great steps forward. Jerry, want to give you the last word. Any cool companies we should be looking at or things that are exciting you without giving away trade secrets? I can't broadcast the companies I want because everyone else is going to choose because there's investments. I don't know, I think I'm going to enjoy spending time actually less with the companies here but a lot with the developers and the customers, right? Because I think by the time that they have a booth here, everybody knows the company exists and it is probably too far along maybe for me to invest, maybe not. But I think talking to developers kind of here, what are their friction points? I think when you hear enough friction either in this ecosystem or another ecosystem or at AWS or VMworld, then there's something there that's got to scratch. Yeah, I was talking to some of the people working the booths and they just said the quality of the attendees here and you just learn something with every single person you talk to and there's only a few shows that say that. I mean, Amazon re-invents one and the quality of attendees always real good, this one and a few others. I think people who come here by definition are all learners, both the companies and the individuals and you've got to want to surround yourself with learners, people who are open, honest, knows learning. All right, well, Jerry, I think that's a perfect note to end it on. We are always learners here in helping help our audience try to understand these technologies. So Jerry Chan, always a pleasure and we'll be back with a wrap up here of day one, DockerCon 2017, you're watching theCUBE.