 Okay, we're back here live inside the Cube Silicon Angles flagship program, we go out to the events and extract the signal from the noise. I'm John Furrier, my co-host this week is Jeff Frick, and we are providing exclusive wall-to-wall three days of coverage here at OpenStack, some of where all the action is here. We normally say we extract the signal from the noise, but in this case, there's a lot of signal, not a lot of noise, a lot of developers, a lot of users, a lot of proof points, and a really exciting open source community. And like other shows we've done, VMworld for instance, we do multi-day coverage and our next guest is Martin Casado, who is the co-founder and CTO of Nasirah, now with VMware, now chief architect at VMware, but your badge says Nasirah. So welcome back to the Cube. Thank you very much. You're loud and proud still on Nasirah. VMware is going through some changes and that's well-documented, we don't want to talk about that, but the action is right now software defined networking, software defined everything. We've actually coined the term software-led infrastructure because networking, servers and storage is all changing, it's still converging, but it's changing in a modern way. So I want to get your take on first, what's happening with Nasirah here at OpenStack, and then we can talk about what's going on other things. Sure, yeah, absolutely, so actually, I gave my first talk at OpenStack in 2010, so this is like a partially packed room where most people were on their laptops and people didn't know what SDN was, and so things have come so far along and we've been such a pivotal part of the community, right? I mean, leading quantum, doing the development on the OpenV switch plugin and we've just been very much involved in the networking and so I'm a huge fan of OpenStack, I'm a huge fan of where it's come and so of course we want to be part of it and so we're back talking about networking and how far we've come along and I actually just walked out of a talk and much of the talk was just a retrospective on how far we've come. So talk about some of the early days because one of the things that we've been highlighting here and why we're here with theCUBE is the community is really exploding fast, accelerating, it's hitting mainstream, DevOps is now a term that's now being kicked around in mainstream, but a term that we've been talking about is infrastructure as code, which is a software paradigm that's really transforming everything and software-driven infrastructure is everything, so talk about where was OpenStack just two years ago relative to the software model because again, it's all open source DNA, right? So just give some perspective and tell us a story or two about the early days and where we are today. Yeah, sure, so like I said in the early days, I mean, OpenStack was kind of this dream and it was kind of funny, like a story I like to tell is Ben Horowitz from Andreessen Horowitz, the VC was on my board and one of the bigger disagreements that we had early on was not how to run the company or whatever, it was whether OpenStack would be viable and I truly believe it was kind of the right model, which is like as a framework for composing best of breed technologies and one of those was networking and that because it was a framework approach and it was a vendor-in-specific approach, it would allow each component to evolve independently and so at the time it was kind of a dream and then we worked on Quantum, so Quantum made networking and- Hold on, I want to interrupt you, sir. So what was the argument with Ben Horowitz again, what was the debate? At the time, he was like, I don't think OpenStack's going to be viable, it's the same as all of these other projects, but what I thought was like, listen, this is actually looking at the problem very pragmatically as a framework where multiple vendors can come and kind of plug in best of breed technologies. So vendor-neutral and choice-driven. That's right, that's right. And attack a multitude of technologies underneath that total framework. That's exactly right and so the focus was early on and I think this is why it's so successful and why it's so great, was to allow vendors to participate and compete on the merits of their technology, which is like I have a solution I want to bring in use and you have a solution and OpenStack abstracts that away from the customer so that they can choose the best of breed, which is exactly the model. So one of the things that we do when we see red flags, we use our blog and we kind of like screen and yell things and editorial and one of the things that we did at OpenStack was because when it started with Rackspace trying to figure out what to do with their servers, prior, I mean their cloud prior to OpenStack, there was a noble mission and we were way behind it but then what happened, it felt like a pool party. Everyone's jumping in, being splashing around a lot of marketing hype, so we were pretty critical of OpenStack as saying, hey, if it becomes a vendor hype environment, then it's going to fail and this track record of a vendor driven nonsense and that's failed. Here, what's happened was, and we've since fallen our sword and said, hey, it worked because of certain things. I want to get your perspective, why did it work? Why did the OpenStack foundation, why did the OpenStack community shield that pool party effect? How did they get rid of those radical viruses that would come in and screw up these great movements and it's certainly they've done that and what was the tactic, what was your view on that? That's such a great question, which is like so many things could have gone wrong but I'll give an example of what went right, which is like I think that they said all the right things early on. I think that Rackspace did a phenomenal job of positioning it and supporting it so I think we should all give huge props to Rackspace for doing that and like seeding control and allowing it so I think that was all great. So we had very good initial conditions but that alone isn't good enough for success, right? So I think the initial conditions were correct. I think the governance model, like of course, there's a, these things are complicated to get right but I think that was pretty good. But then, to allow a company like Nesira at the time to go in and innovate and contribute code and do deployments and not have any pushback and not have any difficulty to allow us to build a subsystem, I think it's a testimony to the fact that not only did they say it initially but they actually followed through. I see a lot of projects, like they will talk openness but then when it comes down to it, it's very difficult to continue to enforce that because you do have a lot of back fighting. So I think that we had good initial conditions. I think Rackspace did a fantastic job of orchestrating that and then there was follow through on this and it's that simple. And they, and people brought code to the table so it became meritocracy on pure performance. That's right. That was one thing that we saw with the user presentations. Real scalable deployments, not some, you know, candied up, you know, proof of concept. Yep, absolutely. So let's talk about infrastructure as code. So obviously you just put out a blog post on the VMware blogs. I think it was this morning, really talking about a fundamental shift in infrastructure. So obviously you guys pioneered STN, you got big switch out there, a variety of others coming in. HP has opened, there's a variety of different approaches. We want to get that in a minute but I want to ask you to explain this fundamental shift that's happening at a technology level around virtualization and an operating environment in this new cloud data center market because this modern infrastructure is upon us and it's accelerating very, very fast, probably faster than we all think it will. I want to get your take on that because a lot of CIOs and a lot of service providers sitting there going, hey I know I have to re-tool, they're starting to kick the tires, we've got to get some new expertise. So share them what's happening and what should they be thinking about as they evaluate OpenStack in context to their modern infrastructure. Absolutely, yeah, yeah. So I'd like to highlight two trends if you don't mind. So the first trend has nothing to do with this area and has nothing to do with VM where it's something that's been happening for a decade which is if you look at modern data centers, say a web 2.0 data center or platform as a service data center, we've already seen much functionality that's traditionally been in the network moving to the X86. These are things like security, things like failover, things like load balancing, things like billing. And now in these environments, the functionality is being implemented at the application layers in the web server, in the load balancer, in the web application. So we're already seeing this broad movement of functionality into software. And this is totally independent of what we're doing. So that's one trend I want to call out. And the second trend I wanted to call out is last year, the number of virtual ports exceeded the number of physical ports globally, physical access ports, just access ports in the data center, which means there's more ports connected to virtual machines than there are to physical machines. And so if you look at those two trends, it's pretty clear that functionality is moving into software and compute's being virtualized, right? And what this means is you have to be comfortable with both of these models. It means that there are these macro level trends that are already happening, which says that functionality will be in software. And networking functionality is going to be on X86 in the server. And so it's just understanding that this trend's happening and then getting comfortable with how you adopt it. And the number one hurdle to adoption once you know you want something is organizational. And so I think that the big thing is know it's happening, understand how to consume it and then start to structure your organization in a way that you can consume it. So let's talk about some of the tactical things around agile infrastructure, as you call it. Or as you use software, you have that agile mentality. In the old days, you had switches and you had configurations. So the normal cloud stack up a bunch of servers, configure some ports and some switches and you're good to go and then you write some software. Now we're moving to an application centric world or a satisfied world where you need the platform, you need the infrastructure. But now software has to make dynamic changes on the fly at scale across a lot of virtual resources. So, okay, how should someone get their mind around that trend and then is that infrastructure as code? Is that just automation? No, that's a super good question. I mean, I think that like you end up having functionality at most layers, right? And so let's look at a modern data center, like a modern web 2.0 data centers. Like you have a lot of functionality already written at the application layer and the application guys do all of this. And so things like billing, like I mentioned, things like security already there in that layer. But there's also the issue of how do you provision a new workload? How do you do an isolation domain? How do you treat five physical servers as one pool of servers? And so that's at a different layer. So at that layer, you use virtualization to solve that problem. And so I think to think at a macro level, what is infrastructure as software? Well, infrastructure as software to me is like doing the provisioning and handling the pools of hardware in an abstracted way. But there's also this problem, and there's also the entire software design space of now how do you build your applications to be easier to use and deploy as well? So again, I think that we're talking about multiple layers here. We're talking about the infrastructure layer, which is something that VMware is very focused on, right? Compute virtualization, network virtualization, storage virtualization. Then you've got the application layer, which also is going to require a lot of work to make it so easy to deploy. And install and move into security and so forth. And so I think, again, this is a fairly complex environment, but it's starting to kind of come into focus. So I'm going to shift gears a little bit to move out of the data center and to the inputs. And one of the big trends we're starting to hear a little bit more about is the internet of things. And as there's more intelligence in things, there's sensors, there's software now on the things, and whether it's your car reporting in at the end of the day or the plane flying across the Pacific that's reporting in the speed of the turbines. Do you see A that is a really huge fundamental shift in terms of the inputs into the internet? And will this cause another great wave of innovation? Or is there already so much stuff going on now that's not that much of a quantum leap? Yeah, that's an awesome question. So, okay, so I'm of the opinion, and I don't know how broadly hell this is. I'm the opinion that networking is almost perfect. It's actually really close. We're really good at scaling the data plane. Here's what networking is really good at. Networking is really good at getting a packet from point A to point B over a complex graph. And whether that graph is really large or whether that graph is really complex, whether that graph is really good at this, that's why the internet is arguably one of the most successful technologies ever. It changed the world, it's scaled, it's fantastic. So networking is very good at that and I think it will continue to be good at that. What it traditionally hasn't been so good at is the operations piece. This has required a lot of people managing state. And so in my opinion, networking is fantastic and it's perfect's the wrong word, but that's almost everything we need to do. Now we just need to solve the operational problem and then we can start really thinking about it. I mean, then we focus on applications. And so the internet of things is very important, but it's important from a consumer standpoint, from a network technology standpoint, in my opinion, we're almost there. And now we just have to make sure that the operations is also flexible. So one of the trends you talk about on your blog post is network functions moving to the edge of the network. So just pivoting off the internet of things question is, okay, that's the edge, right? I mean, the edge is the edge of all things, right? So it could be a phone, it could be a PC, it could be an iPad, it could be a probe. So the workloads are diverse. So, okay, the topology networks, I totally agree that we've had with GoodTrack, we're going to be moving packets around really fast. But now, how does that affect the software aspect of it that you talk about? Because the workloads might be different. It might be a small little trickle of data coming in and all of a sudden, big video comes in, surveillance or other workloads could be a refrigerator. Who knows what the internet of things is. But how does that change or how does your vision of network virtualization? Because you're saying software services at the server level will manage networks. Yeah, yeah, so I think that, A, when it comes to just raw pipes providing bandwidth, the internet and networking does this very well. Which is like, you've got a new device, it's connected, it needs to send a packet to another device, something we're very good. And I don't think that's fundamentally going to change. We're going to use the same distributed types of protocols that converge quickly, that handle change. And I think that's fine. The one piece that's going to require us to work on is how do you manage this? How do you manage the security configurations? How do you manage the provisioning of new devices? And that's exactly what network virtualization is trying to do. And part of that is how do you make sure that there are services on route? How do you make sure that if a new device is connected, that it's got this service available to it? Is that service going to be all the way to the end, or is that service going to be on the network? And honestly, I think we're still trying to figure this out. I think we understand, in the data center, we understand that model pretty well. I think things like wireless and campus were still in earlier days. These are more difficult environments to deploy in because they're brown field, there's not so many green fields in one of those. So I think it's a great question. And I think my answer is, in some areas, I think we understand that pretty well like the data center, in other areas we're still trying to figure it out. Because it's new. Because it's so new, right? That's right, yeah, that's right. So what about the competition? Obviously you guys, and you're dynamic and passionate and at VMworld, you really talked about how you wanted to change the world with this year you did. And SDN's not one of the hottest trends. And we started tweeting. Everyone loves SDN, I don't know what it means yet, but obviously it's a real trend. It's impacting a lot of things and it certainly woke everyone up when that happened. So obviously, big switch was on earlier, HP has open flow. So there's all those different views of what is going to happen. So how do you talk about that and how would you compare and contrast what you're doing vis-a-vis these other opportunities? Because the customers want to decide which way to go. Yeah, no, I think that's totally a fair question, which is, and like when we're a startup, when we're in a Sierra, this is a very important question, like how do you differentiate and so forth? The landscape's so different now. So we're VMware, we've got a large customer base, we've got native integration into ESX and these things. And so we're much, much more about solving customer problems or existing customer problems than worrying about the competition. It's not because we want to be neglectful because there's just so much work to do with the existing customer base. And so when I think about competition, very honestly, it's not about startup X or incumbent Y, it's more about how do you get companies to adopt this technology? So I'd say a primary competition is inertia of the status quo, which is like people are very comfortable doing things a certain way and they don't want to change that. That to me is my number one competition in the field. And by the way, I do hundreds of customer visits. Number one. Number two, organizational hurdles. Like it's very difficult for organizations to consume this stuff. This is really difficult. And then the final one, I just think that there's mass confusion. Like people are so confused on what it means to do SDN, right? And I'll tell you why that is. The reason it is is because we're still in the architectural war. We're still arguing about bits and bytes and we don't have enough products to look at. Like if you're going to test a car, you don't go and argue about the construction of the car, you test the car. And we're finally entering the phase that we actually have products that people can test and deploy. How does OpenStack fit in that? Because honestly, one of the things, people are driving the car. And it's called OpenStack. And so how do you get into that and talk about some of the things that you're doing with OpenStack that's, quote, testing the car, so to speak? Yeah, so I mean, OpenStack is, like I said previously, it's a great framework for testing different types of components. And it's good, it's, you know, customers that want to test new technologies, best-of-breed technologies are standing this up and there's kind of these nice slots that you can slot new functionality in. So it's a great testing ground for things like network virtualization. So it's a very important position in the industry for allowing customers to get their hands on these things. And the cars, it's like, when it comes to SDN, there are very few cars now and they're just starting to come out and get into production. Yeah, so the operational thing is big. I mean, we're hearing the same thing, but there is interest. I mean, the reason why OpenStack is so compelling for the enterprise is that CIOs look at an opportunity to get a warm blanket, to hang their hat on something reliable, to actually do the scale-out open source. So again, I want to ask you that question, scale-out open source versus commercial, software, scale-up? There's a major shift as towards scale-out open source. How do you see that transition happening and what's your view on that whole? So I think there's a little bit of a false dichotomy here, which is, in my experience, customers want to consume technologies in one of two ways. In one of two ways. And the first way is they like, some set of customers like vertically integrated stacks because of the QA matrix, because it's tightly integrated, because of the ease of experience or whatever. I sold against that for four years. There's, I would say the majority of customers, that's how they want to consume it. Not the customers that come here, most likely, but many customers. And so that's, those guys don't want to have the operational overhead of something that you can tweak a lot, and that's fine. I mean, I think that that's one customer base. I think what OpenStack is, is it's focused on the customer base that likes horizontally integrated solutions, that it's not just open source, it's open standards and open interfaces so they can use best-of-breed components. Like, you know, we were huge contributors to OpenStack, Nacero, and we had closed source components because it wasn't about open source, it was about open standards. So if people didn't like our solution, they would replace it with somebody else's solution. So they have software layers. It's a layered base approach. That's exactly right. So OpenStack is phenomenal because it allows horizontal composition and it allows differentiation. And customers love that. Not all customers, but customers do. And it makes sense for VMware to sell to both of these, right? These are both custom needs, but I don't actually view them as competing. Well, I mean, there's shifts of momentum and we pointed out earlier yesterday that, you know, NetApp's not threatened for a long, long time, either as IBM or anyone else. I mean, it's going to be years of changeover. Not, I mean, there's no waking up tomorrow morning and everyone's scale out open source. It's a process. And people who move properly will win. Yeah, there is, but I'm curious to know back to the car analogy, right? So we've got a few of the cars that are out running around and it's, you know, it's 1915 and most people are still on horses. You know, who's Ford? What's the model A that's going to really be the tipping point, do you think, to drive this? The Model T, what's the Model T? Model T, what is the Model T? I wish I knew. Actually, I mean, I think like, what's so exciting about being here now is we're still in the wild West, man. There's like carcasses, you know, on the floor and like people busting out of saloons and like people are like, it's awesome. Okay, Corral baby. I love it. I love the chaos, right? Here we are, right? Like it's not clear who the winner is. It's not going to clear what the model is and is it going to be a services company? It's going to be a software company. I don't know, but like it's going to be important and it's going to be big. And that's great. And that's why we're here. It's because we feel it, right? It's cool. Yeah, we had one comment earlier from HP. It's great to be on the front end of a major paradigm. She had a lot of action, a lot of great action. And you know, there's some chest pumping and going on and some fights going on, but that's natural. Yup. Final question, I know you got to go and thanks for coming on theCUBE. Really appreciate it. Always great to have you on. Is your view of the industry, okay? This is the, take your SDN hat off, put your, you know, observe of the industry. Yeah. Just in general. What do you see out there? Obviously a lot of weird shit happening out there. Yup. And you know, we saw the Boston bombing, changing the world there. It's just weird. Technology's different. You got security hacks. What's your view of the industry, the world around with technology, the future of tech? Wow, that is a very good question and a broad question and one I'm like totally. Come on, take a stab at it. I don't have, but I'm going to grab a piece. Grab a piece. But I'm going to grab a piece of that. It's a buffet. It's a buffet. Pick your favorite. You can go salty. You can go sweet. I'm going to grab the piece that I feel comfortable with, which is, and it's going to be, I'm afraid it's going to be a network and network virtualization, which is. Which is not entirely unexpected. I'll make it good. I'll make it good. So, okay. So here's the evolution I believe of these types of technologies. The first thing you do is you virtualize and you solve like some simple problem. So like that simple problem will be like in the early days of compute virtualization. They're like, instead of buying two servers, you buy one and it's server consolidation. Very simple value proposition. You know, very simple sales pitch and that's why people deployed compute virtualization. And then after that, they started changing the laws of physics and you get vMotion, you get all of this stuff that you didn't imagine early on. But you have to adopt a simple use case before you change the laws of physics. The exact same thing is going to happen to network virtualization. So right now it's like speed to provisioning. It takes some time to get it. But once that happens, we can change the laws of physics because we're already virtualized. And like the analogy I like to make is like, listen, if I was an alien, I wouldn't show you I was an alien right away. I wouldn't be like, I'm an alien. Right? I mean, like, you know, because you couldn't handle it. You'd be like, ah, right. So. That's SDN to the alien. Got it. So instead, you solve a simple use case, you get them comfortably like, listen, I'm an alien. I'm not showing you that I'm an alien, but I just want to let you know. And then after a period of time, you can come out and you're full glory. Martin, great to have you inside the cube. Always entertaining, very informative. We'll see you at VMworld this year. And we're here, exclusive coverage open stacks. I'm a three days wall-to-wall coverage. This is SiliconANGLE. This is what we do at theCUBE. Trek the city from the noise. Make it entertaining and always fun. Martin Casado with the Nacirah VMware now chief architect. Thanks for coming on theCUBE. Appreciate it. We'll be right back with our next guest, Jim Curry from Rackspace. One of the man behind the curtains, this whole movement and we're going to talk to him and just tell stories from 2009 and talk about what happened with OpenStack and should be a great informative interview. So stay tuned and we'll be right back after this short break. Great, thank you.