 Live from San Francisco, California, it's theCUBE at VMworld 2014, brought to you by VMware, Cisco, EMC, HP, and Nutanix. Now here are your hosts, John Furrier and Stu Miniman. Okay, welcome back. We're live in San Francisco, California, VMworld 2012. This is theCUBE. I'm John Furrier and my co-host for this segment is Stu Miniman, analyst at Wikibon.org. We're pleased to have star-studded guest, Martín Casado. Always brings the energy. Day three, we need to get that injection. Formerly of Nasira, which was purchased by VMware and that huge deal everyone talked about. Feels like 20 years ago, two years ago. Welcome back to theCUBE. Martín, great to see you. Hey, it's awesome to be here. Thanks for having me. Looking good doing a selfie in theCUBE. Stu, it's nice. We'll post that later. So I got to ask you, we were just talking prior to coming in. This is, again, another year of SDN. I mean, finally. Well, next year is going to be the year of SDN. Is it going to be this year, next year? This happened in the local area network days back in the 90s, the year of the land. Next year. What's the story? Is it here or not? What's the deal? That's a good question. So I think if I have one high-level point for this VMworld, which is the last seven years, it's coming in the future. It's almost like I feel like I've been leading people, going over the mountains. I'm like, I swear, over that next mountain, that is the time for network virtualization. That's the time for network virtualization. I promise it's there. I know you're tired, and you're hungry, and you have eaten, and you're cold. So the dam broke six months ago, and we're arrived. So now, this is the first VMworld. We've arrived. I mean, and you just look at the commercials, right? I mean, we're at $100 million run rate. We've got wins in every vertical. We've got over 150 paying customers. I mean, it's a big deal. Now this is a real business and a real momentum. It's a shift from this kind of evangelical notion to like, OK, people are using it. Why are they using it? How do you enable sales forces? What does this mean to the channel? I mean, this actually now becomes more of a business issue. Let's go back and rewind. The dam broke six months ago. Just describe that moment. Where were you? What's happening? What's specifically broken? What happened? I mean, what specifically take us through that? Those first couple steps. So maybe I'll provide just a little bit of history, which is OK. So Nacero was founded in 2007. So I actually kind of left my defense, and I went. And I founded Nacero with Scott and Nick. The first version of MVP came out in about 2010. But it didn't have support for vSphere, right? So then in 2012, we got it acquired by VMware. And about nine months ago, did we have a supported version of NSX that supported vSphere that kind of had a lot of the bells and whistles. Since then, we've grown the customer base to over 150 paying customers and $100 million run rate business. And this is in nine months. And by the way, like the core sales team, we just added NSX to the price list for the core sales team in June 6. So that just started. So what we're seeing is we have a viable product to market. We've got a sales team that can sell it. We've got massive customer adoptions and many verticals. And this has all happened in the last nine months. So it's just a tremendous amount of momentum. So I got to ask you, I know Sue's chopping off the bit, but I want to ask you, EMC had the same issue with Extreme I.O. A lot of hype with Flash and just undirected availability. They just, you know, the dump truck of dropping all the products in their existing install base. How much was that VMware starving customers versus product market fit in your opinion? Yeah, no, no, that's actually a great question. So I actually think that we're actually sales limited, right? This is software. We have a direct sales force. These are direct deals. A lot of these are in production. Actually, I was talking with Andrew Lerner recently from Gartner and he talked to 15 of our customer references. 10 of them were in production. So this is a good statistical sampling of them in production. And so I think that there's a real need. There's a real product market fit. And now we just need to find the right use cases to be carried by a sales team into the customer base. So Martine, you know, one of the proof points for me is the FUD in the marketplace is deafening. Everybody out there has said that, you know, the architecture, it doesn't scale. Which, you know, if we look at how this was built, you know, everything that you've done with NSX and not nice here before. I mean, scale is, you know, mentioned in every sentence. So can you kind of address that? Why is that being said out there? And what's the reality? Yeah, I mean, it's so difficult for me to, like, talk to every piece of FUD because, like, there's so much of it. Like, I heard actually recently that NSERA was started in 2004, which is not true, because being the one that started it, it was in 2007. I heard that MVP has been around for seven years, which is not true. I actually contributed a lot of the early code, so I wrote the code and it hasn't been out that long. And so, I mean, they're just so... Can you write the code? Yeah, I was going to say, I'm just saying. Right, so there's just been so many threads and so many narratives by so many people. And when it comes to things like scale and performance, I mean, let's look back. I mean, some of our early customers, by the way, we've never lost an early customer. Some of our early customers were development partnerships. I mean, like, actually, in one customer, I personally remember being there in writing code, I liked to fix stuff while things were happening. So, you know, you have early product in market with customers and you always have issues, whether the scale or performance or stability or whatever. This happens in early code development. And so I think what's happening is I think that people will cherry-pick these examples because it's an architectural issue. Which is like a real classic conflation between, say, implementation and architecture. Like, products have issues, especially early products. But like, architecturally, the scaling is fine. Certainly, performance is fine. And what's nice about this is actually going up the sales ramp, we've got lots of customers using this stuff, and things are looking great. And so, the good news is we don't have to pick historical examples anymore. We don't have to be archeologists. We can look at how people are using it today. Okay, so last year when NSX was launched, we were tripping over themselves to say how they work really closely with you. Can you give us a little bit of insight from an engineering standpoint? You now have the whole, networking is your whole baby. So, you know, what's going on with the partnerships and how are you working together with the ecosystem? Right, so just to be clear, there are sales alliances like resellers and bars and then there's technical partners. Which one do you want me to address? Technical partners. NSX is a platform. We've reaffirmed that with our recent announcements around the Goldilocks Zone, which is a security platform. And so, we're at a point now that we're largely market led, which is we look at the customers and what they want and which partners they want us to work with and that's where we spend engineering resources. Because, I mean, at the end of the day, partnering requires not just engineering resources but it requires us to do validation and design and so forth. And so, we've engaged with a number of partners. We've made a number of large announcements recently with Arista, with Cumulus, with HP. We've done a number of announcements with Paul Walton Networks doing customers with all of these guys, same thing. And so, we engage them as quickly as we can. We do what's best to prioritize resources based on customer demand and we're very excited about the partner ecosystem we've developed. So, even when I hear people talk about the Federation, sometimes people get a little confused. I've heard people say EMC is going to buy a networking switch company or VMware is going to get into the business. The whole project Marvin thing everybody was completely convinced that VMware was going to go into the hardware business. I mean, VMware in its place is software, right? We're a software company. This is a software play. I mean, I think what's happening is, you know, and not just from network virtualization. I mean, you're lifting functionality out of the network and you're re-implementing it in software, whether that software is a generation three application or it's a network overlay. And so, that allows people to view the physical network as basically transport and this is kind of creating somewhat of a food fight where in the value chain are our customers comfortable with? Do they want to do Dell? Do they want to do Dell and Cumulus? Do they want to do Arista? Do they want to do Cisco? And so it's nice to see it open up and go up the market to the side. I think it's way too early for us to choose a winner. All right, so, Martin, first time we had you on theCUBE, you were banging on the table talking about how open source was going to be critical. I heard a lot of open source announcements here at the show. So, yeah, OCP, open stack, and you know, where does kind of, you know, I guess start with open stack and open source. You know, fit into the discussion for your group today. Yeah, definitely. So, I mean, I feel like I'm just like progressively moving up layers in open source. So, you know, we started with Open V-Switch which is a data path layer technology, right? Like, that has been a resounding success and it's in so many clouds, it's adopted by so many companies both competitors and non-competitors are like I still have a large team working on it. It's still, you know, open source and available in Linux. So, I got a question from the community on that. Your, the VMware version of the Open V-Switch, will that inter-operate with other controllers? Well, of course. Of course, yeah. So, I mean, Open V-Switch is Open V-Switch is an open source product that I mean there's, you know, you guys have kind of your own flavor though that ships with NSX. We do not have our own flavor. Okay, it's 100% the same. We have Open V-Switch, we have standard Open V-Switch and Open V-Switch is Open V-Switch, right? And so, that, I mean it is Open V-Switch. So, then you go up there. So, then I say, okay, so the south bound interface, OBSDB, Open Flow, Open V-Switch, something that, you know, my teams have worked on. Then you've got the controller layer. So, you want a northbound interface to that, which is like Neutron and Quantum. I've got two core developers on it. This is a project we've been involved in both cultivating and, or incubating early on and developing. And then now I'm going even the layer above that, which is policy. So, once, once that configuration and management gets worked out, you go to the policy layer and this is the congress project that I started, right? So, also an open-source development effort, a team that I lead that's doing that. So, I mean, I just feel like I'm touching many parts of the stack. I'm just slowly working my way up. We asked Pat that same question so I got to come back to you and bring up a different perspective. I know Stu wanted to go in the weeds there, but I want to take it up a level for the developers. Obviously, the apps are driving and dictating terms to the infrastructure. The apps are hungry to program the infrastructure, right? So, I got to get your take on this. Open-source, obviously, we all love it. And Doc are showing a little bit of a twist on how they're implementing it, which is beautiful and gets really helps developers. That's the container. Okay, that's nice. Great infrastructure as code. The vision has always been get the infrastructure in a position where policy configuration management provision, all the automation can be programmed by apps. Where are we and how the work that you're doing accelerates that dream and that position for the craft developers. Okay, so I'm going to take a totally contrary view here. So, here's my contrary view. I don't think applications that developers want to even think about infrastructure. I don't think they'll ever be an API to the infrastructure. I don't think applications will ever talk to the infrastructure. So, I come back from the days of like... They hate the code, they hate the network, don't they? No, but I don't think they'll ever program it or they're ever going to want to control it ever. So, I come from the days of working and super computers of different architectures. And if I look at the evolution of programming models for the application, before we used to know everything, we used to know cache layouts, the caching hierarchy, page sizes, we know where memory was, what was local, what was remote, we would know different, you know, and if you look at the actual programming model, it's gone far enough a way that now you don't care about these things anymore and you just treat compute like compute and network like network and you don't talk to it. I actually think the history of the computer science industry is people trying to provide APIs to the infrastructure and application writers not using it. I mean this has happened many times in networking. So, I think what's going to happen is application doesn't care about the infrastructure. The infrastructure should be virtualized with a software layer. Totally virtualized as a software layer and they'll be totally decoupled. Like a control plane for the app. So, the app itself has to do its own intelligence. Exactly, the app doesn't care about the infrastructure. It won't talk to the infrastructure, it won't program the infrastructure, the app is the app. It'll be concerned with its data and its computation and an intelligent software layer will manage everything for the infrastructure. Great, let's take that next step. So, what's going on in the plumbing that we see in sequence to pull this off? Yeah, so I think that the most disruptive thing that's happening over the last ten years is we're seeing functionality migrate away from hardware and into a software layer that's an actual infrastructure layer. Now that intelligence, it may be in a hypervisor, it may be in an OS kernel, it may be in a low balancer or a web server, but it's a piece of infrastructure that the app uses and that's intelligence in software as an additional layer. So I think it's valid for us to say there will be a software infrastructure layer. It's a software infrastructure layer, it evaluates its own pace, it provides its own services and the app sits on top of that. And I don't think the app cares about it at all. I think the app is oblivious. It shouldn't care. Exactly, it shouldn't care. So it's the responsibility of this infrastructure software layer to provide a utopian compute surface, a utopian data center surface for the app. It kills DevOps to be called dev. Right? I mean, if what you're saying is true, it's just development. To me, the utopian future is you've got this intelligence software layer that's providing infrastructure and the interface to the user is a policy interface. Now there are always going to be business needs, whether it's security needs, whether it's performance needs, whether it's economic needs, that we have to communicate with that interface. But that's a policy interface. This is policy wonks and business wonks that are talking to this thing. I mean, I always say that the Intel processor never programs it. They have a hard and top on it. So why not create that for infrastructure? Absolutely. I mean, this is where we're going. I think this is the biggest movement in IT in the last 10 years is the movement of functionality from classic hardware into a software infrastructure. How far are we for that utopian? Just peg a year. Pick a solar system where we might fly to to see that. Or in reality, how far along are we on that journey? I mean, obviously it's great vision. No, that's right. I mean, it's great for predictions. In 50 years, will the human race still be the same species? Probably, but will we have the same type of government? No, I mean, it's this decade. How close are we? I mean, our network resolution teases out this concept. I get that. So is it five years? Is it ten years? No, no, great question. So I think for generation three data centers, we've already arrived. We've already arrived. We're already here. We can see it. We can look at it. We can understand the benefits. For traditional IT, we're now going up the adoption ramp. I think we're crossing the chasm. This year is the chasm crossing year, which means we'll be entering a mature market in 2015 on the networking side. And I think we've got a lot of work to do and we'll have a heavy tail to deal with afterwards. Martin, when I hear about applications and infrastructure, I'm thinking about what Docker's doing out there. I'm wondering, I mean, it's early days. At DockerCon, there was like one page written about networking earlier this year. How do containers and Docker fit into your vision of the future? Yeah, so I mean, for me, I'm interested in changing networking. That's what I'm interested in. To change networking, I need to touch as many points in the network as possible. Whether that's a legacy physical workload, it's a bare metal workload, it's a container, it's a VM, and whether that VM is KVM Zen or whatever. And so for me, a container is just another endpoint that I want to connect to. What's interesting about the container discussion though is containers often host Gen3 apps. And so Gen3 apps don't require as much features in the virtual network because they reimplement a lot of it. So one of the core use cases for them is actually security. Because now we can implement security services in a separate trust domain, and we can do that. Fully compatible with things like NSX. So I think that, you know, using containers is great, and that we certainly want to support them with NSX, and we will. So give us the update on the news, your news, on the promotion. Apparently, there's a promotion recently. Is there a new role? I mean, apparently there's I mean, of course the deck chairs always move in VMware, but we do like the senior executive roles. You know, got Pat, you got the back office CFO here and there, and you got Carl, go-to-market M&A, and Bill. What's your new role? What's your official title and new focus? Yeah, so I'm now the GM of the Network Security Business Unit, so I'm responsible for everything. P&L? I'm responsible for P&L, so I want to say that. Welcome to the big leagues, baby. Come to your head. I find that there's lots of synergies in this organization, 1 plus 1 equals 3, and we leverage our core competency to come together. Fingo! How did I do? Well, now you're under a lot of pressure. See, revenue generating. What's your run rate you're doing? You said 100 million? 100 million this year. Just in a... In that a second. Okay, awesome. Well, Martín, always great to get you. I want to get one final bumper sticker. I know one you always have something great to add. This year's VMworld, put the bumper sticker on the show for the folks out there watching. How would you summarize the network virtualization, evolution, and journey? Where are we, and the impact to IT? I think it's two words. Well, one of them is a contraction. Two words, one of which is a contraction. It's here! Okay, we'll be right back. Live, VMworld 2014 is theCUBE. We'll be right back. Thanks.