 Okay, we're back live here in Silicon Valley in Santa Clara, Convention Center. This is the future of networking. This is ONS 2014 Open Networking Summit. This is theCUBE, our flagship program. We go out to the events to extract the signal from the noise. I'm John Furrier, the founder of SiliconANG. I'm joined by my cohost for this segment, Stu Miniman, the chief analyst that we keep on on networking. Our next guest, CUBE alumni, tech athlete, Martin Casada, chief architect in networking at VMware. I wonder if he still has the Nasir of business cards, but great to have you back. Hey, it's a pleasure to be here. You guys are up on the stage. You had like two minutes to talk about the future of SDN Cloud. It felt like two minutes, but the audience was very engaged. Our heads were spinning, some good Q&A. Clearly, one of those sessions where you needed more time, right? People were interested. You're trying to summarize a very complex topic in a very short time frame. So, one, were you amped up? And two, what did you leave out? What have you missed? What could you have done better? What's your take of your own presentation? So, I mean, for me, it's such an exciting topic. And what's exciting to me about is I think over the last year, we've kind of moved from largely kind of theoretical, almost motherhood and apple pie discussions around things like software and openness to, you know, products on the ground and how people are using it. And so my talk was like, wow, you know, we've now built things and people are using them, and it's capturing the imagination. And what does that mean? And so, I mean, I enjoyed talking about that. It turns out that things like network virtualization really do capture the imagination that are being pushed into areas we've never seen before. And so, you know, as I talk about it, I get very excited, and I didn't have the time that I wanted to to go through all of them. But I think like the core message was there. You were talking about things like edges, using slides you were borrowing. It's hard to give a presentation on virtualization without, you know, having all the right pictures in the old router, you know, components, all the old, you know, slideware stuff. So I got to ask you, in our previous talk at VMworld, you talked about creativity, new dawn of, you know, with software virtualizations, SDN. But now we're hearing, this is really the beginning of the infancy of a revolution. We're seeing on the data that we're capturing from the crowd and the commentary DevOps culture of cloud is really a forcing function down the stack and saying, Hey guys, get your act together on the networking side, where you have that DevOps mindset. Now you got operating systems kind of theory at the network level. And then you got what's going on at the virtualization of the network, both logical design principles and then deploying into the cloud. So all those things are happening. One, I want to get your take on those things. And what is the real deal that's happening that that's in virtualization or SDN in particular, that's that's really provable right now. You say, Hey, this that's got traction check. And what are the work areas that need to be improved? Okay, sure. So I'm going to split up my answer into two parts. So the first one, I'm just going to address this issue of like, you know, if you step back, what are the drivers behind it? So I actually think if you look at the SDN space, say this, the SDN space, you've got kind of two drivers. There's a customer use case driven driver, which is basically, I've got business to accomplish. And the business is driving requirements and infrastructure. And I'm going to push that down and require you, the vendors to build stuff to make, let me do my business. Right. And so this is, I think it's the less creative, very practical. I need to get job X done to make money. I manufacture cars. Networking is in the way. This is how that you should be able to change it. And a lot of the discussions around that, and this is how we evolve an industry. But at the same time, there's this kind of very exciting creative chaos that's happening bottom up, which is you get developers and technologists and you put them in a room and you're saying, Listen, you know, actually, forget the business cases, like, we know that they're out there, we know we're going to solve real problems. But like, listen, now we have software, we have abstractions. Why don't you come up with technologies that are that are elegant and general and maybe solve problems that nobody's thought of together. So we got this kind of mix of both worlds. We've got this very practical, let's solve business cases. And then we've got this kind of alternative. Let's be creative and come up with things nobody's coming for. And I think it's kind of a best of both worlds. Do you see the, the incumbents, we were talking on the intro segments, Stu and I, you have the old incumbents, like the Cisco's of the world, the Juniper's EHPs, you have the new incumbents, the VM Wears, and maybe some series D funded startups that may or may not make it. And then you have the startup community, which is now still innovating. So you have three kind of sets of players, right? So the issue is who's going to be disrupted and who's going to be disrupting. So all three players have to make that offensive move right now. So tie your creativity chaos bottoms up into what needs to happen, what's, who's at risk and what's your take on all that, who gets disrupted and who's be disrupting and what does it take to disrupt? You know, I think that is the question. And I've kind of sort of pulled myself out of the prediction game because I think it's very difficult to know. But I do think I can say something concrete, which is a lot of the movement in the SDN space over the last, say 10 years has been about decoupling layers to provide horizontal integration, which means that you have more competition at every level, right? You have competition at the hypervisor space. You have competition at the physical hardware space. And so as you disaggregate, as you have competition in the physical hardware, you know, these food fights occur. And what happens as a result is from a customer standpoint, you get optionality, you get, you get more that you can buy from. But also, you know, prices drop, things become more general. So I think we're going to have this huge food fight. I think there's going to be a lot of collateral. I think there's definitely going to be changed. Like we fill the heat now. But as to who's going to win, I don't know. I mean, I think that the large incumbents, there's a lot they can do with the channel, with the customer shipping product position to maintain position. But if they can't get out of their own way, you know, maybe those that are more established but aren't as new, or maybe one of the new guys will kind of, you know, take the pieces of everybody else wants to be killed each other. So I think it'll be exciting to watch. There will be change, but it's too early to know who's going to win. Stu, before you go there, I want to ask one more thing. So we had open source conversation earlier on with the Chief Science of Brokeade. And yet a quote that said, it's not, it's how you build something is actually more important than what you build. The old days was the product itself. So with open source being such an fundamental role in the raw materials and the creativity, what's your take on that? I mean, the processes and the standards and the openness of that, how you build it versus what you build. You talk about a lot of different things going on with the decoupling and whatnot. Yeah, I mean, open source is such a complicated topic for me. Like I've kind of lost my polygonic view around it, meaning I think you can get in just as much trouble with open source as you can with closed source. I mean, whether it's because you have one person that's committing, whether because whoever adopts it doesn't have the developers or the domain expertise to evolve it, because you have to go through a community to solve business problems. I mean, it's, you know, open source is great. It's great from a community perspective. It's great from a framework perspective, but it's got its own issues. And so, I mean, I think that we're going to live or die by open interfaces, meaning that whether it's open source or not an open source is great and it should always be an option. I think we should all make sure that we're horizontally composable. That is, we should be able to mix and match parts to systems so whoever it is can choose. I want something open source. I want something from this vendor. I want something from this vessel and they can mix and match. So I'm much more now about making sure we have a horizontal composition. We have open interfaces and we stick to those. And then when it comes to open source, I think we have to evaluate that in the case-by-case basis because, I mean, there's a lot of precedence in the industry for it causing as much issues as closed source systems. And so, again, I think it's great. I think it's healthy in certain places, but I think it's a little bit polyanic to assume that we can kind of, it's the only solution for different environments. So, Martín, first time we did an interview with you right after the NYSEER acquisition, I mean, you were banging on the table how open source was going to completely transform networking and change the table here. To be honest, some of the community pushes back and says, VMware absolutely is involved in the open source community, no doubt heavily involved in OpenStack, but to hear your answer now, it sounds definitely like you're pulling back some. So, can you speak to VMware's commitment to the open source community and kind of how you used the statement that you just fit with what VMware strategy is? So, actually, so VMware is actually, I mean, we've doubled the number of developers on every open source project. We've probably contributed more than we've ever contributed before. So, I think VMware as a company is, I mean, Pivotal is all open source. So, I think that there's a lot of work on the open source community. I mean, I think that software and open interfaces are definitely going to change networking for sure. I think open source is going to be an enormous driver behind that. And like, you know, I mean, I've started yet another open source project that's totally open that's called Congress and OpenStack. So, I mean, I keep starting ones. I mean, you know, OpenVswitch, we started early on, Quantum, we started after that. I just started one called Congress. So, I believe that open source is very important. I believe it's a catalyst. And VMware has supported me the entire way, right? Actually, look at the number of lines of code we've committed to OpenStack on all areas. Continue to commit to the Linux kernel over OpenVswitch. Like, the commitment is clearly there. Actually, if you stack this against most other companies, I mean, the contribution's enormous. I just feel that we need to make sure that our conversations are on something that makes it better for the ecosystem. And I think that making it just about open source belies the reality that it's really about composability that gives us freedom. Right? I mean, if I do open source, I can still lock you down on a support model. I can, right? I mean, there's a lot of business things you can do to still make things close. So, just because hardware vendor X does open source or hypervisor vendor X does open source doesn't mean they're necessarily doing the right thing. So, we think we need to make sure everybody sticks to open interfaces. Absolutely. I mean, you know, the attack on open source in general is, you know, open source is free like a puppy. So, you know, you need to be careful there. So, you know, at this show, there's a lot of talk about the various projects. You know, we've talked before about open flow and, you know, that doesn't fit into, you know, VMware strategy so much today. Open Daylight's been a large part of the community here. I know VMware is supportive but not, you know, the largest contributor there. Now that hydrogen's out, you know, what's your take on open daylight and how that fits into the ecosystem? Yeah. So, I think open source controllers are great for SDN in general, especially those that try and solve a number of problems as a base platform. Whether, you know, I mean, I think we released the first one with NOx, whether it's Beacon or Floodlight or Tremor or Open Daylight or whatever, I think this is great. I think the more support and chaos in this, the better the industry is, for sure. But I think it's a mistake to conflate the notion of a general platform with a product. And, like, I've worked on four controllers. I think I built the first two when I was at Stanford and then I worked one at, and two at Nacero. Right? So, I've worked on four different controllers. It's an interesting space. I like working at it. But, like, personally, my arc is now about how do you take these technologies and solve new problems, not how do you focus into the technologies themselves? And so, like, my personal arc is a little beyond how do you build a good controller and more. Now, how do you address different technical problems, whether they're longstanding problems in networking or whether they're longstanding customer problems? I mean, that's kind of where I've gone. So, I'm very supportive. I think it's great that they're out there. I think it's great that there are many that they're out there. But, you know, in general, like, my view is kind of a little bit higher level these days. Sure. This is a personal thing. Yeah, absolutely. And, you know, you're part of the community is not at all. I'm wondering if you can give us the update. What are you seeing with the customers you're talking today? You know, last time we talked, you know, we said it was rather early. It's still the really large companies. If you take this show with this example, you know, you look at AT&T and NTT and, you know, companies with, you know, infrastructure budgets that are over a billion dollars a year. So, you know, of course, that's not the typical company. What are you seeing out there in the adoption of, you know, SDN, network virtualization, and the like? Yeah, so kind of, kind of, kind of taking off my kind of, you know, general GADFLY hat and kind of putting more on a VMware hat, you know, I spent a lot of time in the field and I think kind of, last year was kind of an inflection point. We actually had an earnings call recently where we talked about this where we had a number of wins of an SDN solution which is network virtualization across verticals. And so we had, we had retail. We had large financials. We had telcos. We had Midwest manufacturers. We had airline companies. And so I think that we're getting close to the point of crossing the chasm. And as a technology, so I think that's cool from an industry maturation perspective and a consumability perspective. But as a technologist what I find the most fascinating is as this gets pushed into production it captures the imagination and it's getting pushed into areas to solve problems that certainly I didn't conceive of initially. And that's why I'm so interested in looking at like how does this hit the problem space and solve problems and how are people using it and how can we involve it to make it better do that? So that's where a lot of my focus is these days as a technologist. Talk about what's changed in the landscape. I mean, I'd like to say the high level and the presentation didn't have a lot either but like clouds on of it. So since our first interview two years ago to now, what is, I need to mention your arc has changed and a personal well be written control is now composite apps or composite of APIs or whatnot. But share the folks out there the Martin view of what's changed in the landscape. A lot of action certainly happening. People are finding their swim lanes so to speak. What's your take on? What's changed since all the big Honestly, this is a great question. So I think the battle is no longer about networking architectures. And I don't think the battle is between products or startup. I actually think what we're seeing is a positioning of this arc battle between three it architectures. I think like the entire discourse has up leveled to it. And I think there are three visions that are being espoused. One vision says it not networking all of it should have a hardware basis and it should be vertically integrated. There's another vision that says it should have a software basis and it should be horizontally composed. And I think there's another vision which says it should be only consumed as a service. And I think from a macro industry perspective all of this kind of discussions about networking and storage and software and that has actually been consumed by this mega level grab for IT and the future architecture of all of IT. And so I think this is kind of getting a lot larger than most of us and is becoming the positioning of the titans. And so for me it's kind of exciting to kind of see this. Is it coming from the vendors the customers the developers all three? I think it's largely a stratification in the products available the vendors behind them and the consumption model which is vendors that have margin side to hardware will say IT should be vertically integrated and it should be driven by hardware. Vendors that are primarily a software model say it should be a software defined thing and I should ship it to a software and vendors that provide IT as a service say you know you should consume your data center over your internet link and I will run it for you. And from a customer perspective the macro level messaging is not about you know should it be open source this or should it be you know X project that or should be cloud this it's really how do you consume your IT are you going to buy it as a shrink wrap hardware product or you can write a software you can run on general compute or are you going to use it as a service and that is the battle that's going to play out in the next 10 years. Which everyone's saying it's not going to be one of they're all not usually exclusive but it's all hybrid. That's essentially what we're hearing. In whatever use case you plug it in does that make the argument for composite apps? Yeah and the question is you take the slider bar you slide it out to the future and where's the convergence point and I think some arguments will say it all turns into the Google's and Salesforce's and Amazon's like everybody will consume it from the cloud others will say there will always be a hardware element on-premise and others will say it's actually going to be general purpose hardware over software and I think there will always be a Balkanization but I do believe there will be a true winner. Now it'll be it'll not be a hundred percent market share but I do believe there'll be a monopoly. Well we were just speculating about hyperscale saying you know Facebook's Google and Amazon they built their own because they had smart people who could do it and they say that boatload of cash by doing themselves but you go to an enterprise they might not have the smart people to build it so they'll buy a premium so that's consumption angle. But this is the interesting question so so from an enterprise customer should they buy somebody else's software to build their own cloud or should they just consume the service that they wanted initially as a service right so is it are they building their own data center using somebody else's software or instead of building a data center are they getting that data center as a service and like that's a that's an open question. Yeah absolutely and that's where hybrid fits in of course everybody's doing a little bit of each they've got their SaaS they've got their internal stuff and they've got there and you know that that air gap that we see of course is the apps it's there's some apps that are real easy to move and there's you know and if you're building new apps that you know all the gaming guys are all in the cloud that's right they need to be flexible but it's the mobile guys exactly no exactly but it's such early days I think it's too early to predict what the convergence is going to be right and like there's so many forces on this there's channel forces there's you know sunk assets on customer prem there's sales motion there's there's certification all of these forces are either impeding the movement away or or accelerating the movement array and so these got this very dynamic system that's a feedback loop that we're all trying to predict what it's converge on I think it's early but I think that is the question that is a macro question about IT and the model for IT the thing that we were speculating last at our last event in Vegas was if the consumerization trend happens you've an accelerated traction model meaning because it could be quickly apps fail all the time so the point of failure determines quickly what's accepted so that's one element and the other element is the platform you've got to have an enabling platform so it backs down to the what's under the hood so if you look under the hood of your enterprise or you're networking you've got to have a core stable platform to build on right so that's you can't just say here's my approach and hope it works you can't just you mail that in it doesn't work then everything else fails so you have you know these these forces so the question is to you on the networking side as sdn maps directly in a self-defined data center what is being tooled right now in in real time that's the most critical elements that you can talk to this say to the folks out there not paying attention to the in the weeds the most important elements of of sdn mapping into self-defined data center so I think that there are I think I actually think that the primary model of sdn today is taking intelligence that you would normally put in the network and rewriting applications to consume it and this is not like the traditional model of sdn where you're you have a central controller and you're changing the control model but it's basically saying the network there's not a lot for it to do in the data center especially in the data center and instead you can do it in the applications the problem is is the only way you can realize that architecture which is by far the most successful architecture from a precedence standpoint is if you own the applications right so I think this is very interesting question which is if you can't rewrite your applications how do you get these types of benefits do you buy them from somebody else that can write their applications so instead of running your own applications you buy somebody else's as a service do you go to a company like vmware and you get a translation layer that allows you to run your own applications on existing hardware and it will provide a virtual translation service or do you just continue to run the old infrastructure I mean I think that this is the dynamic I see playing out within the data center I think that the classic view of SDN is like a decoupled control plane I think it's part of implementing those solutions but from like a customer standpoint the network comes down to exactly that question which is do I rewrite my applications do I have a software layer that does a translation for me or do I not change so Martin I'm wondering if you can give us a little bit insight as to now that NSX is out you know how broadly and widely is the networking team inside VMware impacting other products in the VMware portfolio and the federation so you know you'd be cloud hybrid service you mentioned pivotal you know VMware of course has lots of partners you know where you guys you know spending your most time strategic yeah yeah that's a great question so from an organizational perspective we touch everything whether it's VCHS or it's the core or even like VCAC and the consumption model like this is like you don't talk to those EMC guys do you do you we actually X the storage on the EMC side I mean we touch everything and so so all of these discussions go on as far as compatibility as far as integration personally I've actually moved on to doing a lot of security work and so a lot of the work that I do using network as a basis is looking at the problem of security and I'm not sure if you've seen kind of recent stuff I've done on the Goldilocks but I actually think there's a whole new effort and attempt to kind of revolutionize the security space in a very similar way and so that's kind of my you know one of the areas that I'm personally very invested in yeah yeah I mean your keynote this morning you said that 40% of the customers that were buying network virtualization were doing it because of security and honestly I would not have predicted this even though like that was the genesis of the work I did at Stanford was security actually you look at my first paper it was a security architecture like it turns out like this is why people are adopting it and so like this is kind of like I'm personally very invested these days into the security space talk about the Goldilocks zone okay if you could share your opinion on that yeah okay so what is the Goldilocks zone so so here's the rough idea so if you look at the security industry security spend actually outpaces IT spend as a trend and the only thing outpacing security spend is security losses and so I think there's an architectural issue and the way to describe that architectural issue is today when we push security controls out there we either get rich context or we get isolation but we never get both so if we put security controls in the application we get all the context we need we get who's logged on and what data's being accessed and how it's being accessed but you're in the trust zone of the application so a malware can turn it off if you put security controls in the infrastructure you're very well isolated you're in a different attack or a different trust zone but you have no context and so the Goldilocks zone says actually the hypervisor is in this very unique position where it's close enough to the application to have context but it's far enough away to actually be to be isolated so you kind of get the best of both worlds and so I'm working on a large platform initiative to enable partners to use that layer to provide both context and isolation in the security product so the term was coined in NASA in the 70s to describe the simultaneous things that need to be in place to sustain life on a planet not too hot not too cold you know it's a balance it's an ecosystem ultimate ecosystem life but in a way that comes back down to the platform viability issues where there's so many things in place it's not about one or two dimensions there's a lot of different elements so interesting so where are you in the project you wrote a blog post two days ago about this or I think yesterday about this what was where is the stage of this project right so so the ultimate idea is to enable the security ecosystem to build on a platform to imbue their products to have rich context and isolated enforcement and this cuts across kind of you know all areas well many areas of security whether it's vulnerability assessment or it's next generation firewall or it's anti-virus or it's network access control I mean like any of these will actually benefit from this type of approach and so we've actually have a number of systems that are in production today built on this as proof points we've got some pretty core technology that's been developed over the last year and we're generalizing the architecture going forward so this is far more than a blog post at this point there's actually a lot of proof points behind it but it's still it's early about generalizing this and really enabling the ecosystem so that's kind of where I'm working on now we're getting short on time here but I want to get that out there I mean security obviously a big part of it you heard services so and that's one a philosophy but with cloud and all these under the hood examples there's still a lot more innovation take the golden locks projects as an example huge innovations I always always ask you on the cube here what's your view for entrepreneurs you've been a startup entrepreneur coming out of Stanford very well depth and a domain expertise folks out there that are watching this infancy of a revolution of massive change great innovation is upon us what would you share the folks out there would be your view on what to look at what to sink your teeth into what to pay attention to just share the share the love well I'm you know I mean it's something that I think about a lot clearly and and and these days I spent a lot of time actually exploring the startup ecosystem from a partnering perspective and a strategic perspective from within VMware and so much and like chaos is opportunity and there's a lot of chaos and and ergo we've got a lot of opportunity but so much of infrastructures just about insertion and so as you're exploring interest and as you're exploring these changes make sure that you understand how technology is inserted not just technically in the stack but at a market in an organizational level because without that you can build the best technology in the plane you can take advantage of all the buzz and the hype but you really can't get these things deployed and at the end of the day I think that that's as important as core core technology yeah so Martín along that line so I'm wondering you know talking to all the customers that are looking at this technology is there something they can do to move this forward faster or something that's you know the you know is it just inertia is it you know the organizational structure you know how how can the user community you know move this way faster honestly I think at this point it's as simple as just talking more I don't think it's about I mean I think that companies are dealing within natural laws of organizational physics and I don't think there's anything you can do to catalyze that as far as like pushing without breaking something however if they open up they talk more they're more willing to describe their internal use cases which and there's so many of them that I'm directly involved in I think that you get more of a kind of a group thing and a comfort around this and so I think be open participate in the community and share your internal use cases Martín final question because we got a break just quickly summarize for the folks out there why is this show so important why are we here what is the future of networking why does it matter what will accelerate just quickly share your quick summary of what why this moment in time is so important yeah I mean I think we're definitely in an inflection point in the industry and there are so many vectors that pushed on it whether it was the mega data centers moving things in the application or virtualization consuming ports into the software layer or open flow catalyzing the discussion about decoupling the control plane or open stack creating a network there's so many vectors that have pushed on the industry it's cracking and when it cracks change is going to happen and so it's incumbent on us as a community to get together to provide direction to minimize the impact and provide guidance because like listen the future like is like okay this is a cliche but like really this is like this a future is ours to shape we've got this in co-op mass of shattered networks that was created by all of these vectors and now we need to provide some level of guidance so like we all need to be here and we all need to participate and we all need to be very clear about where we're going and we should do it great opportunity infancy of a revolution whole new dawn of great stuff happening if you're an old incumbent new incumbent or new entrant great opportunity to be a disruptor great change this is theCUBE we're documenting and continue in our fifth year now thanks for coming on again great to have you always great to have Martina sharing his knowledge great candid tech athlete this is theCUBE we'll be right back with our next guest I'm John Furrier with Stu Miniman we'll be right back after this short break