 Okay, we're back here live at VMworld 2013. We're live in San Francisco, California, the Moscone Center for VMworld 2013. This is theCUBE, our flagship program. We go out to the events, extract the seal of the noise. I'm John Furrier, the founder of Silicon Amgen. My co-host for this segment is Stu Miniman of wikibon.org. And we have here Kelly Harrells, the Vice President and General Manager of Software Networking, Business Unit, Brocade. Welcome to theCUBE. Thanks for having me. So obviously the network virtualization, software-defined networking, has essentially created software-defined data centers. So we were just talking with Dave Vellante about the software-defined data center. It's kind of the destination, the modern era. And the question is, what leads? The storage? Now you have software-defined storage out there. You have software-defined networking. Tell us your perspective, because obviously networking has been an era that's been innovative lately and enables a lot of change. So talk about what's happening with that in context to VMworld. Sure, well it's easy actually to follow the pattern on it. Just follow the app. Follow the app and you'll figure it out. Anytime there's a major architectural change to IT, it starts with the application. Nobody goes and changes their network and waits for an application need to fill it. Just change the application model and then the storage model has to adapt to that. And then the networking has to race to catch up. And it's always been like that. If you think about from mainframe to client server, this transition happened. You had to change the storage model and then change the network. Same thing, client server to web server and then now in cloud. So we've got the application model changing, storage followed, and now networking is really racing to catch up. You know, it's always, I have a friend who's a CFO and he always says, John it's so easy managing companies, just follow the money. So if you're a CIO following the app as an interesting perspective, but that's a new change that's now just been thrust upon infrastructure providers. I mean, just what is the impact from your perspective? I mean, this is, you know, you go back a decade or further, you build the networks, a few apps you support. Now as a tsunami of apps and the apps are dictating policy and all kinds of other things, how is that disrupted? Give a sense for the level of disruption. Well, it's massive because what's happening with the apps, it's not just that the apps are virtualizing, they're centralizing, this is the cloud model, right? And they're sharing resources. So what this is causing is it's causing the need for more and more of the functionality that you find in traditional IT architectures to start merging and blending in with the server. So, you know, in networking, for example, you've got the V-switch, that was the first step in. Now you've got routing and you've got security and you've got distributed models. You've got the similar impact on storage as well. So the big challenge, you do follow the app, but the app, virtualized app world is just exploding right now. Last stat that I saw from Gartner said that more than 50% of the x86 servers that are shipping are shipping virtualized. And not only that, but those servers are bigger and they've got bigger and faster Knicks in them as well. So the whole app model is just fundamentally changing. What pressure is that putting on the network? I mean, the network has been a bottleneck in the past, but now it's exploding with SDN. What are some of the specific things that are happening at the network level? Well, there's a couple of things. One is how you manipulate the network and the other is what's being manipulated, right? So how you manipulate the network, this is really where the broad brush strokes of SDN come into play. This is the idea that you've got separation of controllers and forwarding plans. You can feed logic into a controller and it gets distributed out to the forwarding plans to execute the desired behavior. Now that is already underway and there's a lot of different ways to approach that. The second is what's being manipulated and it's not all hardware anymore. Now you're also manipulating networking software, networking virtual machines, virtual routers, virtual firewalls, this sort of thing. So you've got kind of the breadth of what you're manipulating and how you're manipulating it and that is what's causing all of the innovation and networking right now. Yeah, so Kelly, in the keynote this morning, Pat Gelsinger said that networking is going to be really one of the big focuses of the show. They announced their NSX and they used a term. They said, we're going to do a radical evolution. So it really struck me as VMware is now a mainstay in most enterprise IT companies and if I look at Brocade, Brocade's a mainstay. I mean Fiber Channel is in a lot of environments but you came in through an acquisition. You were the CEO of Viata. You've been doing really software-defined solutions before they were called software-defined. Very hipster of you. So can you talk to us, does that messaging of radical evolution translate with to you and what have you been seeing in your transition into Brocade? Well, thanks for the ping on being hip. So hip I can barely see over my own pelvis. I think the, I got one in the back office there. So the radical evolution concept is actually, it's a necessary requirement because people are not ripping and replacing their networks. You just don't do that. Networks evolve through a succession of upgrades, right? And the thing that actually enables this particular radical revolution is, or radical evolution I guess as Pat said, is the fact that you can extend the networking into the server. You're actually adding new networking capability. You're not ripping old pieces out and putting new pieces in. You can still have your ethernet, the switching fabric, and add L3 through 7 functionality into the server. So you can do that and you can do it quickly without a massive boil the ocean approach to the re-architecting of the network. Yeah, so really if we look at what VMware did when they put their layer of abstraction in, it actually had that ripple effect to disrupt how storage and networking were managed. Cisco long ago came out with a 1000V to try to solve that by saying Cisco should own everything and manage it like they did. You know, explain to us what this new announcement for Anisek means. Brocade is part of a large ecosystem, just about every networking company, except for Cisco, who is absent there, was in there. What does this mean to kind of the management of the physical and virtual network? Well, first of all, you have to have both. You have to have hardware and you have to have the software, right? And a good example of the interfacing between those two is one of the announcements that Brocade is making at the show here is in our Brocade ethernet fabric is a VXLang gateway capability that really seamlessly bridges NSX with the ethernet fabric. So you've got to have them both and they both need to be able to interoperate very, very well. I think the bigger picture of what's happening in the industry, you don't get very many massive technology disruptions in your career in IT. This is one of the biggest that you'll ever see because all rules are off. The idea that everything has to be a bigger, faster, better box and that it's what we did the last two years but just slightly faster, that's off the chain, right? Right now we've got cloud architectures are spreading applications all over the place. We didn't have VM density before. Virtual machine density, you've got big multi-core servers, 50 virtual machines in each one. How do you manage the traffic between those? How do you manage the bandwidth explosion per rack of servers that's going on there? There's brand new approaches need to be taken. So one of the questions that comes up in the ecosystem discussion is DeFi convention is the theme here at VMware and past five years there's been more change in the previous five years. You look at what's happened, Moritz came in, he's now got his pivotal spun out. You saw the HP performance was on just unit shipments, the old model, you got Steve Ballmer's now been resigning from Microsoft. So a new guard is coming in a new way and this all kind of happened. So with that DeFi convention, what is the core thing in the ecosystem for VMware? I mean overlays, underlays, at that network virtualization is a lot of innovation, a lot of bets are being made. What is the outlook? I mean, your perspective is you're in that market, is it underlays, overlays, separated, they're together, either, both, what's your take on that? Right tool for the right job. There's going to be both underlays and overlays. I think you're going to see the overlay model really pick up the most in the cloud and host types of environments. And then the underlay will be more specific first to some of the type of private clouds. But I think that those two are going to be converging over time, but you need both overlay and underlay depending on the problem that you're trying to solve. If you're going after a certain application class, if it's a very latency sensitive trading application, the approach that you might use might be very, very different than the approach that you would use if you are a cloud service provider and your objective is to scale out your architecture to be able to handle thousands and thousands of tenants. Kelly, one of the challenges I see for networking is when I talk to the general practitioner out there, people that are helping consult, customers aren't asking for SDN. And so much of it is what's real, what's shipping. And one of the things about your solution is it's shipping. You've got, I believe you said, of the free download version, there's like a million of them out there. And I think the number, if I remember, it was over 10,000 paying customers. So can you walk us through, what was the pain point? What's the usage and as you're maybe a little bit more on the NFV side than what we consider SDN, but what is that discussion and how does the industry need to move forward to get customers to buy into this whole next wave of networking? Yeah, the free download model that we originate our business with basically put us in the position of being like Visa wherever you want to be, right? Because well over a million right now around the world. But the problem that people were trying to solve and the reason that they began commercially adopting Viata was simply the fact that I have many virtual machines in a single server and I've got racks and racks that are replicating this. And I have got to segment and secure my network that we're using a lot of the same techniques that I'd used before, but it's not working from a hardware perspective. I need to be able to reach into the server and get those same type of controls that we had before. And again, the NFV model is a very, very simple one. It basically says that we can take the power of an Intel server and with a hypervisor on it, we can run network virtual machines on that server the same way that we would run application virtual machines and we can manage the traffic between those and I can get rid of all of my compliance and concerns that otherwise I would have and I can get rid of a lot of performance issues from tromboning in and out of the server. So it's the thing that has driven the adoption of NFV and during the adoption of Viata in that capacity has been that requirement. Now the customer segment that was the first to really adopt it was the cloud service providers and the biggest ones and Viata is live in Amazon live in Rackspace and in clouds around the world. And it was precisely the cloud service providers desire to solve that segmentation and security problem at a pertinent basis that caused the adoption of the Viata virtual machine and then now with the telco's interest really exploiting on the scene is saying we want a lot of those advantages too. Now the NFV term has been coined and it's good for cloud, it's good for telco, it's good for enterprise. What's the approach that you think, to your perspective, actually there's a lot of pets being made and I was just talking about some of the introduction about the religion of the politics, Republicans and Democrats of SDN. I don't know which one fits into which camp but obviously open source has been driving a lot of innovation and open source is kind of a new standards body. So we've thought they've seen open stack out there has been enabled and Nassir is playing early on in that before they were acquired by VMware. But you've got the big whales coming in like Cisco with a different approach, other people are trying things differently. Still the jury's not out yet on the approach. Is the software free and you charge for the hardware or you give the hardware free and charge for the software. With all you can eat, one flat price or you kind of lighten it up and charge for software upgrades. So a couple different approaches going on right now and so each one has its own benefits. What do you think is the right way or what do you think is going to happen? I think it's very customer specific. So I think that if you are a cloud service provider what you need is you need to be able to spin up your services as fast as you can with as little upfront money as possible and actually pay on a usage-based model. So you've got that kind of model, you've got the traditional model where somebody wants to acquire it all upfront and then put, you know, capitalize it and then you've got the services and support model as well. So the nice thing about software is that it gives you incredible range of how you can choose to package that. So you can actually model for business. There is no silver bullet. Now, you know, if I presented you with one model for your business, I could turn right around over here and Stu might say, well, wait a minute, that doesn't fit my business and I can change it very, very easily to match his as well. Kelly, I wonder if you could speak to the kind of the network administrator. You know, so much of the discussion is how can we get from this very manual world to some kind of automated environment and what does that mean for kind of the future of the network admin? Matt Gelsinger said, you know, the VMware admins that are here, you know, you are the, I think he said, you know, mobile cloud, you know, person of the future. I'm not sure what that term means, but you know, we know that roles are going to change. You know, what are you seeing? What advice do you give to practitioners out there? First thing is look up. Look up, look above the CLI, right? You know, get up to the rest levels. Get up to the orchestration levels because increasingly that layer of abstraction is going to keep maturing in the interface between those orchestration levels and the actual infrastructure underneath it that's executing commands is going to give you a lot more leverage. But you know, the CLIs are not going away overnight, but what is happening is the desire for more of a button click and get a response type of an environment. And that means that open APIs are going to be crucial to this. RESTful models and orchestration that could be anywhere from the VMware orchestration model to an open source orchestration model and then all the way down to other proprietary alternatives. So I guess I've gotten pushed back and say it's not about CLI versus GUI because CLI, you can actually script some things much better than say a GUI where you had to click the remedial button. That's right, that's right. You know, where is the maturity of that? You know, the RESTful APIs and the orchestration today? It varies. In our case, it's very mature. As a matter of fact, down in the show floor right now, our demo is using a server with VMware ESX on it, applications, the auto virtual router, and the database that those applications need to access is actually up in the Amazon cloud where there are also the auto virtual routers. So through a few button clicks through RESTful API, we're able to establish secure connections so that the applications on the server can access the database in the cloud. And that's all through the REST API. So there's varying degrees of maturity in it right now, but from our perspective, we just assumed that there was a different adoption model which is why we led with that kind of an interface. I've got one last question for you. I've worked with Brocade since 2000 back in the early days of SAAN. Most people that I know still think of Brocade as a fiber channel company. What's your viewpoint? Haven't come in and haven't worked there. What is Brocade's position in the marketplace today and how should people think of them? And I know you're not the marketing guy. Well, you know, there's, you don't want to be basically myopic about it. The simple fact is Brocade is the world's number two vendor of data center network infrastructure. And that started with SAAN. We have Ethernet fabrics. We have IP routing. We have a variety of products that are in that portfolio. So if you back up and talk about data center network infrastructure, Brocade is the number two vendor and we're in a position right now to really begin exercising that strength aggressively in the IP networking space. Kelly, we got to wrap it up, wrap the segment up. I want to ask you one final question to the last word. Okay, the business is changing. A lot of folks in the trenches are looking for navigation on kind of where the wind's going to blow so they can get the wind up their back. What do you say to the IT guys, to the guys in the trenches, guys who are looking at gear of history and experience up into the CI. What's the sound bite for them? What should they be looking at? What should they be thinking about in this new era of NFV and virtualization and SDN? What should we be looking out for? Well, they have tremendous skills in knowing how networks are built and how to manage them. They just have to transfer that skill into the new arena, which is that it's also happening inside the server. So within that, from that perspective, it's not scary, don't be afraid, jump into it. There are some new skills to develop though and the leading edge architectures are going to be the ones that are starting from the server and working their way out. So get ahead of that curve. Moving to the server, you're seeing it with flash, certainly obviously the network being smart with virtualization makes the network scalable and enables a lot of growth for the applications. Follow the applications. Kelly, thanks for coming outside the queue. We'll be right back with our next guest here. VMworld 2013, day one of three days of wall-to-wall coverage. We'll be right back after this short break.