 Live from Santa Clara, California, it's theCUBE, covering Open Networking Summit 2017. Brought to you by the Linux Foundation. Hey, welcome back everybody. Jeff Frick here with theCUBE, along with Scott Rainovich. We're at the Open Networking Summit 2017. The Linux Foundation has taken over the show a couple years ago, it's a lot of excitement and a lot of people would say that the networking was kind of the last piece of the puzzle to get software defined, to get open. And we're really excited to kick off the show with a really great representative of SDN and everything that it represents. Barton Casado, now with Andreessen Horowitz. Martin, great to see you. Hey, I'm super happy to be here. So, coming off your keynote, you said it was 10 years ago, almost to the day that you guys started the adventure called Nassira, which kind of put us where we are now. You know, you and I are growing old together here. Like, it has been a decade and I've actually been on theCUBE throughout, so I'm very happy to be here. Thanks so much for the interview. Absolutely. So, what were your takeaways, Scott, on that keynote? It was great. I mean, we had some great stuff this morning. Not only was Martin given the history of Nassira and the origins of SDN and talking about how you made it successful after all these kind of challenges, but we also had AT&T unveiling a new incredible white box program where they're running, you know, open networking on their entire network now. So, it was kind of a, I thought, a big day in general to show how far we've gone, right? You talked a little bit about that. Yeah, yeah, and actually, I mean, listen, having kind of come, you know, every year since the inception of ONS, what strikes me is, I mean, it originally, it was so speculative. It was kind of like, wouldn't it be nice? And you had all these dreamers. It was, you know, largely academics or people from the CTO's office. And if you compare like those first meetings to now, like, we're in the industry proper now, right? I mean, if you come and you look around, like this huge representation from telcos, from vendors, from customers and academics, so I think we've seen a massive maturation in general. I just think we have so much, I can make a mashup of all the time if you've had you on theCUBE today. It's coming, and we're almost here. And they're like, it's here! But now Tom Donovan said that their goal, I don't know if it's in the short term or the very near term, is to be over 50% software defined. So I guess that's a pretty good definition of being here. Yeah, yeah, I think so. I mean, I think that we're seeing, and I think that the HAT talk was fantastic, but I think you're seeing this across the industry, which is, you know, large customers that have been traditionally conservative have these targets, and they're actually implementing. I mean, it's one thing to have something on the roadmap, and it's one thing to have something planned, it's another thing to actually start seeing it roll out. And so, I mean, again, this is a process. I mean, a lot of my talk was like, you know, how long does it take for an industry to mature? But like, now there's many things you can point to that are very real. I think that was one great example of it. Well, the other thing I thought was great in your talk is you kind of mapped out the 10-year journey, and you said, you know, it's so discounts often, the hardest part, which is changing behavior of the market. That's much harder than the technology and some of the other pieces. Right, exactly. So take this from a technologist standpoint. Like, I basically made a career on making fun of hardware. I'm like, software is so much faster than hardware, and hardware is so slow. But now, by stand back, and I take a long view, like, yeah, fine, hardware's slower than software, but it's nothing compared to changing like organizational behavior or consumer behavior. And so, I mean, for me, it was actually pretty humbling going through this last decade because you realize that like, even if you have product market fit, and even if you have a good technical solution, there's a natural law of market physics that you have to kind of overcome a moment of inertia that takes probably a decade, I mean, certainly five or six years. And that's before things like vendor viability when you're trying to enter the enterprise space or legacy infrastructure, which is just not getting ripped out, so many hurdles. Strictly consumer behavior, right? Like, strictly consumers are used to doing one thing. I mean, I always talk to new entrepreneurs and I say the following. You kind of have two jobs as an entrepreneur. Job number one is you identify a constituency. That constituency wakes up. They think about everything in the world, but they don't think about your thing. So job number one is again to think about your thing. That's difficult. It's like inception. It's like Leonardo DiCaprio inception. You're like, you're like putting an idea in somebody's head. And then the second thing that you have to do is you have to attach a value to that. So just because they have the idea doesn't mean that they actually value it. And so you actually have to say, listen, actually this is worth X amount of dollars. And it turns out that this takes a long time. And that's why market category creation is such an effort. And that's why it's so neat. We're standing here and we're seeing that this has actually happened, which is fantastic. You talked about Nacera, which I mean today, correct me if I'm wrong, it's still kind of the biggest success story in SDN in terms of eight startup, 1.3 billion. You talked about different iterations. I think you said six or seven product iterations and being frustrated at many levels. Did you ever sit there one day and think we're going to fail? Like, was failure? Oh man, I don't think there wasn't a quarter. Oh, and we're like, we're dead. By the way, that's every startup. I mean, I'm on six or seven boards right now. I mean, every startup kind of has this oscillator. And when we started at Nacera, it was in 2007. And then 2008, the nuclear winter setting, if you remember, right? I mean, it's like the whole economy collapsed. And I think that alone could have killed us. So absolutely, and all startups, you do that, but one thing that I never lost faith in, was that the problem was real. I wasn't sure if we had the right solution or the right approach. We iterated on that, but I knew there was a real problem here. And with that as kind of a guiding star and a guiding light, we just kept going towards that. I think that's why ultimately we ended up solving the problem we set out to. It was just, we took a very crooked path to get there. What was the feedback mechanism? Was it like just talking to as many customers or as possible or talking to them? You talked about the market fit versus the industry fit. How did you gather that information? I think in core technical infrastructure, the strategic leaders of a startup have to be piped into the nervous system of both the technology trends and the product market fit. Technology trends because, I mean, you know, technology trends provide the momentum for what's going to get adopted and what it looks like. And the product market fit is, what is the customer problems that need to be solved? And so I think it's really critical to be deeply into both of those things, which is why things like ONS are so important because they do kind of kind of convergence of both of that, what the customers need, but also where is the technology going? And it's really neat that it's kind of like the platform versus the application. You're going down a new platform strategy, right? Which is the software to find networking, but at the end of the day, people buy solutions to problems that need to get fixed today. No one's buying a new platform today. Yeah, yeah, yeah. So I think, well, okay, so there's two issues. There's definitely, yeah, you're right. There's the technical directions and then the specific applications for that. And one thing I talked about and I really believe is like we focus a little bit too much on the technology platform, how those are shifting early on and less on what the customers need. I don't think you want to 100% flip that. You need to focus on both, but I think that they're both should be even heavier. Exactly right, like what do customers need and then what is the right technical approach to that? And you also touched on a couple of really interesting points about kind of decisions, right? You're going to make a lot of mistakes going down the road, but she said there's really two, you got to make two or three really good ones. And that will make up for a whole lot of little missteps along the path. So in retrospect, and this was actually a big aha for me and maybe it's obvious to other people, but this is a big aha for me. Even as I was putting together this talk. So the way venture capital works is you make a lot of bets, but only one in 10 will actually produce returns. And so you're kind of swinging for the fences and almost all the returns comes from the Googles and the Facebooks and the Ubers and so forth. That's just how it is. Now as a venture capitalist, you can have a portfolio. You can place 10 of those bets in parallel. Going back through kind of all the slides and everything we've done, I hadn't realized before how similar doing a startup is, which is like, you can make a lot of mistakes in startups, but a few key decisions really drive the strategy. Does that make sense? It's like, I always thought like maybe you need to do 50, 50 or maybe even 80, 20, 80% correct and 20 wrong, but it's not that. If there's a few key decisions that make it correct. And so the key is, is you're straddling these two pieces of human nature. On one side, you want to stick with something, you want to make sure that like you're not sticking too long with something that isn't going to work. And then the other side, you don't want to stick with, you don't want to get rid of something before it's going to work, right? So you want to be both honest with yourself when it's not working, and you want to be patient. And if you do that long enough, I think that you will find what are the critical decisions to drive the startup forward. Yeah, one interesting thing you said, you arrived at a conclusion that the products and individual applications were more important than the platform. And that kind of runs contrary to the meme that you have now where the Harvard Business Review is saying, oh, I'll build a platform, build the next Airbnb. And what you're saying is kind of contrary to that. Right, so I went into this with a platform, I mean, if you look at our original slide deck, which I showed, it was a platform. Now, I think that there's two aspects for this. I think in SDN specifically, there's a reason technically why a platform doesn't work. And the reason for that is, is networking is about distributed state management, which is very specific to application. So it's hard for a platform to reduce that. So technically, I think there's a reason for that. From a startup perspective, customers don't buy platforms, customers buy products. I think if you focus on the product you build a viable business, and then for stickiness, you turn that into a platform. But most customers don't know what to do with a platform because there's no necessarily value add. So products before platforms, I think is a pretty good adage to live by. But design your product with a platform kind of point of view that way so you can make that switch when that day comes. And now you're just adding kind of applications, applications. So I want to shift gears a little bit, just kind of about open source and O&S specifically. You know, we hear time and time again about how open source is such an unbelievable driver of innovation. Think of kind of how your story might have changed if there was an active, maybe there was, I wasn't there, something here. And how does kind of an open source foundation help drive the faster growth of this space? Okay, great, yeah. So I actually think, I'm probably in the minority of this, but I've always thought that like, open source does not tend to innovation. That's not like the value of open source's innovation. If you look at most successful open source projects, traditionally, they've actually entered mature markets. Linux entered UNIX, which is a, so I'd say the innovation was UNIX, not Linux. I would say Android went into like, Palm and Blackberry and iPhone. I would say MySQL went into Oracle. And so I think the power and beauty of open source is more on the proliferation of technology and more on the customer adoption and less on innovation. But what it's doing is it's driving probably the biggest shift in buying that we've ever seen in IT. So IT's a four trillion dollar market. It's this massive market. And right now, in order to sell something, you pretty much have to make an open source or offer it as a service. And the people that buy open source, they do it very different than you traditionally do it. It allows them to get educated on it. It allows it to use them. They get a community as part of it. And that shift from like a traditional direct vendor model to that model means a lot of new entrants can come in and offer new things. And so I think it's very important to have open source. I think it's changing the way people buy things. I think building communities like this is a very critical thing to do. But I do think it's more about go to market and actually less about innovation. So what does it mean for all these proprietary networking vendors? I mean, are they dead now? No, no, no. So here's actually another really interesting thing, which is like, I think customers these days like to buy things open source or as a service. Those are the two consumption models. Now, for shipping software, I think shipping close source software, I think those days are over or they're coming to the end. Like that's done, but customers will view whether it's on-prem or off-prem and appliance is a service. So like, let's say I create, you know, Martin Hub. So it's my online service, Martin Hub. People like Martin Hub. Now I can sell them that on-premise. Now Martin Hub can be totally closed source, right? Like Amazon is totally closed source, right? But people still consume it because it's a service they think it's open. And then if they want something on-prem, I can deploy that and they still consume it as a service. And so I think the proprietary vendors need to move from shipping close source software to offering a service. But I think that service can just be on-prem. And I think we're seeing this shift happen. So I don't think there's going to be like a massive changing of the guard. I do think we're going to see new entrance. I think we're going to see a shift in the market share. But like it's not, you know, this isn't like a thermonuclear detonation that's going to kill the dinosaurs. Yeah, yeah. I'm going to give you a take, Martin, on kind of the next big wave that we're seeing, which is 5G. And really 5G is an enabler for IoT. So you've been playing in this space for a while. As you kind of see this next thing, getting ready to crest. What are some of your thoughts also sitting in a VC cherry price? You all kinds of people looking to take advantage of this thing. That's funny. So I'm actually going to answer a different question. Which is, as all the speakers do. Because 5G doesn't exist yet, right? No, I love the question. But it's like, this is like really a space that's really near and dear to my heart, which is cellular. And I've actually started looking at it personally. And I mean, even in the United States alone, there's something like 20 million people that are under connected. And I think the only practical way to connect them is to use cellular. And so I've been looking at this problem for about a year. I've actually created a nonprofit in it that brings cellular connectivity to indigenous communities, like Native American tribes and so forth. And- As the ultimate last mile. As the ultimate last mile. So which is interesting. 5G is fantastic. But if you look at the devices available to these people to coverage, I think LTE is actually sufficient. So what I'm excited about, and I'm sorry about answering a different question, but it's just such a critical point. What I'm excited about is, it used to be $150,000 to set up a cell tower. Using SDN, I can set up an LTE cell tower for about $5,000. And I can use existing fiber at schools as backhaul. So I think now we have these viable deployment models that are relatively cheap that we can actually connect kind of the underprivileged with. And I don't think it's about the next new cellular technology. I think it's actually SDN's impact on the existing one. And that's an area, of course, I'm very personal in it. I love it. I mean, it is, as you said, it's repackaging stuff in a slightly different way over the technology to do a new solution. And it's truly SDN, right? I mean, if you look at this, there's an LTE stack all in software running on proprietary hardware, I mean, sorry, on general purpose hardware that's actually being controlled from Amazon. And again, a factor of 10 reduction in the price to set up a cell tower. Awesome. What about the opportunity with internet of things and connecting the things with networks and artificial intelligence? So as a venture capitalist, when it comes to networking, I'm interested in two areas. One area is networking moving from the machine connecting machines to connecting APIs. So we're moving up a layer. So you've got microservices, now we need a network to connect those. And there are different types of endpoints. And they require different types of connectivity. But I'm also interested in networks moving out. So it used to be connecting a bunch of machines, but now there's all these new problem domains. Like the internet is moving out to interact with the physical world. It's driving cars. It's doing manufacturing. It's doing mining. It's doing forestry. And as we reach out to these more mature industries and different deployment environments, we have to rethink the type of networks to build. So that's definitely an area that I'm looking at from the start. What kind of activities there? I mean, you have guys coming in every day, pitching new automated connect the car software. I think for me it's the most exciting time in IT, right? It's like the last say 10, 15 years of the internet has been the world wide web. Which is kind of information processing. It's information and information out. But because of recent advances in sensors due to the cell phone, the ubiquity of cell phones, the recent advances in AI, the recent advances in robotics, that internet is now growing like hands and eyes and ears. And it's manipulating the physical world. Any industry that's out there, whether it's driving, whether it's farming, is now being automated. And so we see all of the above. People are coming in. I mean, they're changing the way we eat food. They're changing the way we drive cars. They're changing the way we fly airplanes. And so it's almost like IT is like the new control layer for the world. Right, right. All right, Martin, thanks again for stopping by. Unfortunately, we had to leave it there. We could go all day, I'm sure. And come up with more good questions, Wayne. All right, well, I really appreciate you taking the time. It's good to see both of you. Absolutely. All right, he's Martinez Otto from... Andreso DeHorowitz, I'm Jeff Frick, along with Scott Rainovich. You're watching theCUBE from Open Networking Summit 2017. We'll be back after this short break. Thanks for watching. Okay.