 Live from Seattle, Washington. It's theCUBE, covering DockerCon 2016. Brought to you by Docker. Now, here are your hosts, John Furrier and Brian Graceley. Okay, welcome back everyone. We are here live in Seattle, Washington for DockerCon 2016. This is SiliconANGLE Media's theCUBE. It's our flagship program where we go out to the events and extract the signal from noise. I'm John Furrier with my co-host Brian Graceley. Our next guest is Avinash Lasaksham, who is the co-founder of HeadVig, or founder of HeadVig, and the founder of Cassandra and Dynamo, former Facebook CEO. I was one of the inventors of Dynamo. One of the inventors, founder of Facebook. He's also known in circles. No, he's been to Facebook and Amazon. Welcome to theCUBE. Thank you, thank you. You're kind of a celebrity in tech circles. Certainly a lot of experience. Really going through the early days of the web scale, we call the pre DevOps movement where you're kind of making your own stuff at scale. Now we got a little bit easier environment for developers. First question is, what's it like now? Look back then and now, what's changed the most in your mind from the early days of the DevOps battles? You know, first off, thanks a lot for having me. And for me, primarily, no matter what the titles get associated with me, I've always been thinking and like an engineer. I mean, if you see me in the office, I'll be one with the boys doing stuff. So for me, it really didn't change that much mentally. And I think the training I got was more at Amazon where when we built Dynamo and in Amazon, I believe it exists even today, that the service owners are also the guys who are responsible for it operationally. So we had major duties and you get called in when there are outages or not and you got to figure everything out. There was no build something and hand it off to like a separate DevOps. You build it, you run it? Exactly, yeah. Exactly. That's the way I kind of like it because when you build systems, you always go with a theoretical understanding of how you would solve the problem. But once you start running in a reasonable amount of scale and you know, Amazon kind of puts it right out there and later Facebook, even with a larger scale. But the point being that you got to run these things operationally for you to understand how you can retrofit your designs to make your life easy, right? Making your life easy is the key to make things run operationally very efficiently. And at those days, look at the Amazon, a bookstore if you will, now Amazon web service, $10 billion business, Andy Jassy's kicking names and kicking ass over there. And now you got Facebook, which is where you work all applications in and of themselves. That's the norm now, people are moving towards that world where it's an application focus, then they dictate infrastructure and software. That's the modern era we're living in. That's what you guys have based your business on. Yeah, see, for me, depending just like how things in life are relative, right? One can never claim he's the smartest. One can never claim he's a dumbest either because you're going to have a guy around the corner who's smarter than you are or dumber than you are. If you look at applications, depending on which level of the stack you fall in, the guy above is an application, the guy below is not, right? So it just depends on, to a guy who builds the operating system or if he's used to working in the kernel, you might be storage running in the user space, but you look like an app to him or her, right? So it's all relative in a sense. It's kind of where you, your view, where you are in the stack. Exactly. How do you look at the trend of the horizontally scalable infrastructure? Because commodity hardware and cloud creates a horizontally scalable architecture, but yet apps are becoming purpose built towards the top of the stack. So you've got big data, kind of pre-package specialization, domain expertise, big data analytics. Yet the commodity scalable, horizontally scalable is the real value of some of the cloud stuff. How do you, from a stack standpoint, what do customers do to deal with that? Used to be easy, just build a stack and that's your app. See, I think fundamentally, hardware has come a long way in the last decade, right? And if you look at why people have started moving towards these horizontally scalable apps, the reason for that is that, if you go back, say, 20 years, why did sand come to be, right? People were creating islands of storage and it was becoming a nightmare to manage, right? So people came up with this idea of a centralized storage infrastructure. One person could manage it and all apps kind of get storage coming out of that environment, right? It was a good design, worked well for a long time, right? But if you look at how things are moving now, all applications are implicitly horizontal in nature, right? You take any database. It's not a monolithic piece of software. You take HBase or Cassandra or MongoDB or DynamoDB. They're all kind of scalable in nature, horizontally, right? If you now, what people end up doing is now creating islands of storage because you cannot put them on a traditional sand or a traditional mass, right? Because all the scale-out properties are lost when you go to a centralized storage infrastructure, right? Now, I came from that world where I built highly scalable, horizontally scalable kind of database-like applications, right? But what I learned was, if you look at what happens in storage, it's been no fundamental innovation in the last decade. There have been good companies, good exits, but they've all been kind of incremental in nature. Well, pure storage went public. So to argue, they're still overvalued, some say, but since NetApp, that's the only one that's kind of like a matter of the woodwork. Exactly, but if you take these scalable applications, they're creating islands of storage because of the monolithic storage environments that are on the market. So what's the answer? What if you could take that sand and make it a horizontally scalable quote-unquote sand, right? Be it sand or whether you, what protocol you expose doesn't matter, but as long as it can scale horizontally, now you could take the same idea of a sand from like 10, 20 years ago and apply it in a modern setting, right? But you got to keep costs under control, too, and you want to keep things more flexible. So you want to run this off the shelf. I mean, do the users, I mean, the customers care as long as they get the data, right? I think there's a really, there's a subtle nuance here. So what, I think what happens is people hear things like, well, you know, I came from Facebook background or I came from Amazon background and a lot of enterprise companies go, well, we're not like them. You know, they don't understand the applications. But what happens, I think, is, and we talked about this a ton, the key to these next types of things is you've got to take operational costs out of the system. Exactly. You've got to make that simple. And so what you're doing, we've seen Nutanix do this a little bit. They used to talk about, hey, we have all this Facebook or Google DNA. The customer doesn't have to see any of that. They want to see a block storage, a file system. And what you're giving them is essentially the DNA that you had at Facebook, the DNA you had at Amazon and going, don't worry about it, but this will make it simpler, it will make it cheaper. Is that where you're going with that? That is exactly the way we want to do it, right? And if you, if the reason why almost every customer we have spoken to, almost every single one of them has a mandate from above to look at alternative architectures for storage. It's kind of unprecedented, right? Even three years ago when I was out fundraising, something that I don't really like to do, but you know, you always get asked this. I was always getting this question. Storage will not move really fast. People are risk averse, et cetera, et cetera. But now that question has kind of been taken out of the equation because everyone has an alternative. A look at an alternative. The growth in data alone has been pretty massive. The availability of data is the new currency. Exactly. And so again, the premise, and then the developer growth is incredible. So make the data available, horizontal, and make the developer productive on the app. Exactly. Not the plumbing. And we have taken this, I think, a couple of steps further and I believe, I'm willing to stand corrected, but I don't think apart from us, anyone is even close to doing this, which is we can span our cluster across different data centers, spread geographically, no matter how far apart, right? So we have a customer that's basically running our software spread across four or five data centers spread across two countries, right? And it's one cluster that spans that. So you get DR kind of out of the box just by clicking a few buttons from a provisioning perspective. You can push that analogy even further. One of the data centers could be the public cloud since we are all software based. You could spin our software instances on AWS and you can have an on-prem and AWS cluster kind of living on the same physical cluster, right? What's the big trending conversations that you've had with customers? As they look at re-architecting and re-platforming their enterprises, they want faster development, we hear that all the time, but when they look at their overall architecture of their environment, what are the top conversations that you see? I think it's the simplicity of getting something up and running, like how you would do things in an AWS-like environment, right? You go there, you don't wait for weeks for VNs to be provisioned, and most of the time, the storage provisioning that takes up most of the time. If you go to AWS, it's a few clicks of the button here and there and everything is done for you. And on top of that, the kind of policies that we allow one to enable on these storage assets that we give out to applications like cross-data center replication, pick and choose which data centers you want the data to reside in, et cetera, et cetera. And being able to do that literally through your iPhone, I think it's a very potent premise for folks. So when you were fundraising, what was the biggest challenge you had when you were going out? Because you were, back then, storage wasn't that obvious. A lot of people were nervous, as I remember those days. This NetApp was just kind of like, not performing well, and EMC was sideways, and... You know, fortunately, I was kind of lucky in the sense that people were willing to take the bet on my past experience, right? And almost every round that I've raised so far was preemptively done rather than me actively going and seeking. But still, I think... It bet on you. That was approach, and it was your approach too. You were betting that the world would move. Exactly, and the writing was, at least in my mind, on the wall, there are certain predictions that I made that I will never make on a public setting like this, and they were all spot-on a year later, right? And the trend that you see right now, if you look at what we set out to do and what we are doing now, we have been doing this for about four years. So make a prediction, come on. We never pivoted. So whatever it is we are doing is a way to move forward. And I think what's going to happen going forward is that... See, I think ultimately there will be this journey that goes to the public cloud, right? And you will have people who would want to run across multiple public clouds, right? Because you don't want this vendor locking, right? If you go into one particular cloud environment, and if you start consuming the services that they have, you're locked in. Your apps are locked into that environment. Now you get squeezed for price, et cetera, et cetera. It's easy to move compute than it is to move data. So we believe, and we have this capability today, which is if you could build a universal data plane that can reside across multiple cloud vendors, then you can move your compute really nearly, still get locality of access for data. I think that's where things will ultimately start shifting. And IoT fits nicely into that. Yes. What about compliance? That comes up a lot. We hear, well, I love the organic sprawl of the data, the sand kind of gets spread out. I love the elastic resources, horizontally scalable, you know, focused applications. Then it's like, someone goes, damn, I can't forget the compliance question. Or in that regard, security. Exactly. So I mean, compliance and security kind of, it's a very big subject, but I can't give a one-line crisp answer to how that problem is. Just say yes. Yes, we got it covered. But compliance is a little bit more of a blocking and tackling, securities could be technical, right? Exactly. I mean, some of the customers have asked us about whether we do encryption or not on the data that we store on our systems. There's multiple ways you can skim that cat, but some people may or may not kind of embrace that. You could use SEDs, like self-encrypting drives, and you could say that's the first step towards kind of providing encryption. Because the biggest problem with encryption is who does the damn key management, right? I mean, do we do it? I don't want to be in the business of doing that. But if every vendor has their own mechanism for key management, then there has to be some kind of an API that we can embrace so that we can drive that from our system. So the formula is, it's got to be software. You've got to be able to distribute your data anywhere because you don't know who wants it or gets it. It's got to have an API. And behind the scenes, it's got to have some pretty kick-ass technology to make all that stuff as simple as possible to use it. Absolutely. I mean, for the first time, I'm sure a lot of people do this now, but we always claim that provisioning should be so brain-dead simple that you should be able to provision thousands of virtual disks out of your environment for your VMs or bare metal or container environments. And it should be as simple as being able to do that through an iPhone. And we believe we bring that to the market. How many employees do you guys have now? Well, less than a hundred right now. So you're still small? Still small. So small. Still hiring, though. Still hiring? What are you looking for? For hires. From an engineering background? Yeah. I mean, mostly engineering. R&D is a primary focus tool. I mean, I'm looking for people who are extremely passionate about doing large-scale distributed systems. If they have the background, great. Different people have a different view on distributed systems. Not all the views kind of gel with my view of what distributed systems should be. The most popular view is a piece of software runs on multiple machines. It's a distributed system. Not in wine. But even if people don't have that background, if they have the willingness to learn and if they have a good systems background, more than happy to make that investment. Yeah, so you're like systems programmers. Founder-led, founder-driven. And you love those kind of companies. I love founder-led companies, believe me. Because when the market changes, you've got to have that kind of gut feeling, the best you're making. As you said, you've got to know the market. Final question for the folks watching. Just explain the Docker madness that's here because show's growing, leaps and bounds. Very similar to AWS in the early days. Rancher labs, containers here. Mesos, CoreOS, Docker, Kubernetes. It's kind of noisy. How would you kind of break it down for someone who's not here? What's the big walk-away? Where does it all fit? Well, I wish you could give me the answer to that one. We're working on it. But from our perspective, I think for us, it's an integration with all these various pieces of orchestration software that are available. We will go to bed with any one of them. I mean, today we have- You're under the hood. You're just infrastructure pool for them. We made three big announcements today. One is our integration with ContainerX. They are the v-sphere of containers. They bridge the gap between running containers in a VM environment, Linux or Windows. We did this whole announcement with Flocker. They provide persistent container volumes and the whole integration with Docker data center. Which is, I believe if it picks up, that will be the orchestration tool of choice. But they will be given a run for their money from Kubernetes because it's backed by Google and Google- But that battle's being fought above you. Exactly. So you're under the hood, you win either way because the word compose, composable DevOps fits right into your wheelhouse. Exactly. I mean, it's, you know, there's a quote that I've read and I try to use it whenever I can. So guys who know me know that I'm going to try to sneak that in. Yeah. You know, it's- They were bad if you're going to get it in or not. Victor Hugo said, no army can withstand the strength of an idea whose time has come. I think what we are doing, the time is now. So it's for us to fail. And I think if we execute, we could be the number one in the space. Yeah. And I think that's the great chart. Thanks so much for sharing the insights here in theCUBE. Really appreciate it. Congratulations. Thank you. On your startup, Hedvig is doing really extremely well. We are the co-founder. We'll be back right back with more live coverage from DockerCon. I'm John Furrier with Brian Gracely. You're watching theCUBE.