 We were back live at VMworld 2012. I'm John Furrier, the founder of SiliconANGLE.com. This is theCUBE, our flagship telecast. We go out to the events and extract the signal from the noise. And we're excited day three. We're kicking it off. I'm here with my co-host Dave Vellante, founder of Wikibon.org. And our guest is Dean Steve Herrod from VMware CTO. Busy guy out in the field last night, AT&T Park with NetApp. Talking to all the partners, great keynote. CUBE alumni, welcome back. Thanks. My fourth VMworld on theCUBE, or is it my third? I think third or fourth VMworld. We've had you on before in other venues, I believe. That's right, structure. That's great. How do you feel? I mean, obviously VMware, one of the things we were commenting on at the end of day one when we were wrapping up was, VMware is moving very, very fast beyond just looking at VMware. And people who know VMware know that's always been the focus. VMware and moving that together and going back two years, Paul Merritt's announced it, you know, the map and the roadmap and laid that out. Now this year, it's bigger. Bigger vision, a lot more going on around that and orbiting around VMware ecosystem conversations. Pat Gelsinger, you're doing a lot of meetings with partners, so tell us what's the vibe like out in the ecosystem now that we're two years out post the big announcement from Paul. What's the ecosystem like from your perspective? Yeah, I really enjoyed this event in particular. Obviously larger on all fronts, 20,000 attendees, which is quite fun. And I think my favorite time always at these events is going to the solutions exchange where you go out and you see the different partners, what they're doing and how they're integrating. And we certainly saw it here again. What's also very nice, I think we've helped the industry paint a vision of where things should go with the Software Defined Data Center, which you've heard a lot of people talking about. And I think each vendor recognizes that's how things are headed and they're looking for how they contribute in that new vision. And you can, again, you can just see it in the announcements that have come out and then the solutions exchange demonstrations. So one of the things we want to talk about, Dave and I want to drill down on is obviously Software Defined Networking. This year announcement was the big event that's happened recently. And it rivals kind of like the spring source in terms of the impact to what it means. Obviously it's a lot more in terms of dollars. I think Jay Shree from Arista called it the Instagram for networking. Big huge acquisition. But first, let's talk about data, data infrastructure. So Dave and I are introducing this concept that we like called data infrastructure where converged infrastructure is a great vision, it's been happening, but now data is in the center of the value proposition and it's changing some of the architectural components. We see that with Flash. So what's your vision around data infrastructure or data-led infrastructure these days? Because there's a lot of on-premise reconstruction going on around architectures, different use cases, depending upon the company, that's got to go work with clouds. How are you seeing this data-centric infrastructure? That's a great area to discuss. So we definitely see that we can make the Software Defined Data Center be unique in the history of IT and that it can be one platform that really works well for your existing applications, but it has to be relevant and help out the new type of application architectures coming together. And I think if you look at the past of IT, it shows up that you've created infrastructure for a specific application need. That in turn has led to silos which have led to a lot of inefficiency. So we're particularly interested in these data-generated architectures is if you go out right now, you might see people beginning to experiment with them and they're creating a proof of concept as a new standalone part of an infrastructure. So these are silos that have not yet been invented. And so we really want to get in the middle and avoid them there. So towards that end, there's some things you really need to do in this new Software Defined Data Center towards them. And we announced a couple of things around flash recognition. And we're beginning to treat direct-attached storage as a first-class citizen in this new environment specifically to get the bandwidth and the things you need for these new applications. Okay, we said storage was sexy back in EMC World 2010 with Joe Tucci. Remember, and he's like, You couldn't get him to agree, but we coined that term. You said storage, we used to be called Snorage about now, storage in the center of the value proposition. At least data, right? And obviously big data is the big buzzword that really talks about mobility and apps, et cetera, at the top of the stack and user experience. Question I have for you is, you talk to the community a lot, obviously the technical community and VMware's got a huge community that's growing. What's the impact of the systems that are built around Software Defined Data Center networking? What's the impact those systems will have on the new environment that you see? When you talk to people, what are they saying to you? Like, wow, this is going to be really revolutionary in this way. Can you share just some of the things that anecdotally you heard, or things that you specifically were involved in where these new systems are going to have an impact and specifically areas that you think that are going to be relevant? It's fun because I think it's both technology impact and it's also operational impact. And we're seeing this every one we speak with. Technology-wise, it's certainly about how do we abstract out this infrastructure in new ways? How do we pull it together so that you can more easily dole it out? And then the most important reason for all of this to automate more than you could in the past and thus move faster and move less expensively. So everyone likes that technology shift and again you'll see all the hardware vendors looking at that angle, you'll see software vendors looking at it. But the harder part I think in many cases is how do I as an existing IT department prepare myself for this new area? Some more on the technology side but I think one of the more impactful things we did at this event is launch a cloud operations set of instructions. And these are things like how organizations change when you get to a successful model, how the different silos you have in your org might need to collapse. Even a fair amount of IP around process changes. So I think it is kind of neat because you have to change both to get the full benefit of this new approach. So I'm curious, Steve, it's your point of view on the following. So I understand the operational impact. I'd like to drill down to the business impact a little bit. Two things strike me. One is you're just talking about more flash and becoming more flash aware. And the other is Hadoop, the way you're reaching out to the Hadoop ecosystem. One of the things that the data-led infrastructure premise that we have is that big data analytics will feed transactional applications in near real time. If we can define real time as before you lose the customer. And that requires new data infrastructure, maybe flash led, where the point of control is really the server or the VMware really to do things in near real time. So will that lead to, first of all, is that premise, do you agree with that? And will that lead to business productivity impacts that are dramatic? I love this area, it's a great one to discuss. We briefly, on our first keynote, only showed a real smidgen of what we were thinking the world looks like as we head forward. And we showed a shared architecture in kind of a contrived fashion where you had web traffic going on, people doing typical purchases or whatever it might be. And on that exact same infrastructure, we were able to deploy and run Hadoop. We just showed a little bit of it, but can deploy it quickly. And we're actually using the same infrastructure to have more analytics going on at nighttime, more web traffic, for instance, during daytime. But it's kind of neat if you can get to that same architecture, you can very quickly move back and forth between them and the data is in one place so that you can't operate in near real time. And especially for kind of these web analytics, we're able to go in and say, I can give you recommendations or that sort of thing. Real time is before you lose a customer for most of these cases. So I think if you can share the infrastructure for the two types of data, you really get a benefit there. Okay, and I wonder if we could come back to the flash piece. PolyNIST calls it the horrible storage stack. PolyNIST of Intel. Really, she's talking about the spinning disk, but not only the spinning disk, the things like SCSI protocols were very chatty. Do you envision the day where we can get to a point of atomic rights and maybe two orders of magnitude, greater application performance, again, getting back to the point of driving business productivity beyond just operational productivity and doing more with less? Absolutely, and I think, again, there are some great announcements on the floor here of everyone doing all flash arrays to things that are plugging into the PCI slots. And I think about it right now as another layer in the hierarchy of speed. So you have on-processor caches, you have memory, now we have a new rung, which is flash, and then you have traditional storage and archival storage. And if you really think about this continuum, then people are gonna find ways to get the things that you access most frequently into those layers. And that's what we launched an initiative called VFlash, virtual flash, which tries to treat flash memory on servers just as we're treating memory and CPUs today and really let people dole it out quickly and leverage it. But I think it is a profound change on not just the storage industry, but just overall, as you said, application performance and even architectures too. Steve, I want to ask you about Nasir, I talked about server-defined networking and server-defined data center. Obviously, that got everyone's attention. Great messaging, which by the way, I totally believe in the data center of the future being very software, very big data-driven. And we've been following it since 2007 in terms of things like probes in the network and all kinds of kind of new operating system, if you will, new operating environment. So kudos to that, we're 100% behind you on that as a vision. But right now, server-defined networking is what people know around Nasir, but has the word networking in it, right? So data infrastructure, conversion, storage and compute. So server-defined data center, take us through your vision, connect the dots between server-defined networking, a la network virtualization, and server-defined data center, and bridge that and share the folks that roadmap because this confusion is server-defined networking, networking or data center, and how does that fit into this data infrastructure that we're promoting in the future? Yeah, I think it fits very well into what you're discussing. And it's a little bit ironic because I think that the CPU and storage side of the world has moved a bit more towards software-defined abstractions than network is yet. But I think the reason so many partners and so much of the ecosystem is looking at the notion of software-defined networking is because it is one of the more rigid aspects of a data center today. And it really is one of the things that keeps you from really making your whole data center be- You can say bottleneck. I don't know about bottleneck, but certainly the complexity and a lot of the rigidness, it ultimately doesn't let you automate as much as you need to. And so I think that's why there's been such a big recognition of it. So specifically, software-defined data center is a way of pooling all of the resources of your data center together and using automation to dole it out as needed. Software-defined networking is one of the most important pieces that remains to really do a good job on to allow that vision to happen. So I would say it's a subset of the overall vision, but one of the most critical ones, and obviously that's why we're very interested both in our own products, as well as using NYSERA to help us branch out and provide those same constructs across others. So where are we on the maturity level, in your opinion, Steve? I know this is subjective, but if you say on a scale of one to 10 for abstract pool and automate servers and memory, let's say a 10, nine, whatever, it's pretty good. Where would you put storage? Where would you put networking? And where would you put security? Yeah, great question. In fact, I've tried to do that in a couple of areas that we've looked at. I think compute and memory, I think we're at a, I'll give us letter grades instead. I think we're at a solid A. Plenty more work that we're doing and can do, but I think we're with B motion, everything else we're in great shape. I think storage is at a solid B, maybe a B plus, with some of the new abstractions coming out, some things that let you pool storage together and move it around. There's definitely still some restrictions that we need to knock down on that front. We did one with enhanced B motion, for instance, which lets us move things around even if they don't have the same sand beneath them. And I think networking is where we're in the very, maybe a C plus type of area right now. There's definitely great stuff that's been done, but too many cases right now, you still know the topology of how your network is laid out. You still have to worry about subnets or VLANs that don't have enough scalability. And those are quite restrictive. And the good news is that, again, across the whole industry, people are recognizing that and going after it. But in many ways, it is the furthest behind, but it's also one of the most impactful if you get it right. Steve, there's a nice story in Wired Magazine around Nacir Acquisition 70 engineers. I mean, they kind of got it right, but it was a good story. You've got some new talent coming in. So my question is twofold. First, I want to talk about this acquisition of talent in the sense that it's not just Nacir right now, although they're software-defined networking and that's going to create some enablement across those other areas, compute and storage, as you mentioned, and put that together with your partner, stitch it all together. I see that's a cool vision. But I want to talk about DynamicOps. DynamicOps is a really sweet acquisition, the sense that it's kind of like no one was talking about it, but it's really kind of strategic because what they do is simplify a lot of things. So the question I have for you is one, where does DynamicOps fit in the stack? Is it on top of vCloud? Is it underneath vCloud? Because I can make the argument that sitting on top of vCloud is a better end user piece of it to enable and simplify abstract, if you will. Do you agree with that? Or are you guys talking about that internally? And what's your vision on that? Yeah, great question. We're really just getting now into talking about it publicly. So first of all, when we go to our customers, we very carefully look at how far along in the journey towards a real cloud model they are. And we notice that a lot of our enterprise customers who had moved very far along the journey, were using DynamicOps. So obviously that sparked our interest. And really, in short, what it's able to do is recognize the workflows and the things you have to plug into when you're making real enterprise decisions over where something runs. And it ties that with offering out self-service to employees who want to use, they just want to run their applications. Not to the help desk. I mean, for the folks out there, DynamicOps basically is a service layer, if you will, for lack of a better description, I'm always simplifying, but it allows direct end-user or edge-of-the-network management and services that are basically abstracting away some of the details. Yeah, Gartner calls this new category a service governor, which basically means it's trying to choose which services based on compliance rules, based on cost in these different areas. So to your question, John, it absolutely is the very top layer of the stack for us. It is helping customers make the right decision, letting employees move quickly. And then it ultimately will say, here is this virtual data center created by VMware. Here's another one. Here's our data center in a different state. Here's Amazon. Here's physical hardware. So it's a very top-level view of possible places to deploy your applications. It doesn't do a lot after it hands off. It lets those handle it. But it's very key for making the right decision and letting people offer self-service in a safe way. Well, thanks for clearing that up because I think that was an exciting announcement that was kind of unsung in terms of, well, I mean, people noticed it obviously, but this year it takes all the headlines. I want to ask you kind of some geeky questions that we're following. Obviously, you're a software guy, you've been around the block and you're CTO. So at the bottom of the stack at the low level, there's a lot of work being done in the computer science area around low-level virtual machines. Compiler, kind of a compiler discipline. So take me through innovations as an example at the lower end of the stack. Take us through your mind what's happening relative to some research and innovations that are moving up and down the stack relative to virtualization. Could you share some insight, your opinion there? Obviously, the lower end of the stack is one example. Yeah, it's a fun area. And maybe I'll shift more to the end user computing side of the world for just a moment to talk about that. You know, we really talked about for the first time the notion of multiple container types, meaning there's a lot of abstractions that are important ways of doing really what virtualization about, which is separating a logical view of something from how it's implemented. We've actually done this in the past with core virtual machines, but then we bought ThinApp, for those aware of it. It's a way to abstract out an application from its ties to Windows. We announced some interesting things on the mobile phone space where we're actually doing full virtualization that lets you run an entire Android stack on it. But we also announced something on the iOS side of things. It wraps up an application. And so I think the main point I really like is that there are a lot of constructs that are important. Virtual machines I think are a great one that solve a ton of problems, but there are other ones as well. And I think our challenge as an industry is be able to choose the right abstraction for the right use case, but provide a way that you can manage all of those different ones. And it's a much longer, great discussion, but that's really what we're trying to do on the horizon suite side of what we're doing. Let me ask a different way. So abstract, pool, and automate obviously is kind of codes for operating system, which we love. What are the highlights in the abstraction area that you can point to right now that you think are super cool that people may or may not be aware of? Yeah, so two big things on the abstraction front. The biggest is as you raised earlier, which is the network. People are recognizing you can separate out how you view logical networks, logical wires, logical Knicks from physically how it's done. Second very exciting area is the L4 through 7 services that live in the network. You can think about these logically and then offload the execution of them to either software or hardware. And then lastly, I just wanted to reiterate, we're really recognizing direct attached storage and server flash as two new things coming into the system. And if we abstract those in an interesting way, everybody can get more out of those investments that you're making as well. It's always great to have Steve Herrod, CTO of VMware on theCUBE, and just we love talking ESPN of tech, as we say, and shooting the breeze. Great guests. I guess my next question is just, okay, we've been talking about this modern era here in theCUBE and on SiliconANGLE, Wikibon. We are in a new modern era. You guys put out the big slide old way, a new way, PCs, mobile, apps, big data, servers, cloud, great, we love that. What happens to the old stuff? So what happens to all those apps that are the legacy apps in this new software defined, automate pool, abstract pool, automate? What happens to the old stuff? Do you throw it away? Does it work? What's this technical strategy there? You know, this is where you ditch the technical idealism and you really go to how businesses work right now. Every company that I meet with is looking at an inventory of all their applications, and they're doing a form of triage. These we wanna, we can't touch them, we can't change this. We're gonna wrap them up and make them work as well as possible. These we're gonna invest in, invest in them to make them more scalable, to have a mobile access to them, these different areas, and these we're gonna completely rewrite for the new world, and every company I go is creating these buckets and putting their applications into them and then adjusting accordingly. We're getting the hook here, but I wanna ask one final question we always with Steve, love to have the final question be more of a shoot the arrow forward for the next year. What's on the to-do list for Steve Herrod and the Office of the CTO in terms of figuring out, one, operationally fill in the big white spaces that need to be filled in, if any, and two, now that Paul Moritz is gonna be the strategist getting a new canvas, if you will, to paint that next generation picture with Joe Tucci. Talk about those two things from your standpoint because you gotta straddle both. So what's on your agenda on those two fronts, operationally in terms of objectives and then painting that new canvas? Yeah, great way to close up our discussion. So it is really two things. We had a report card. There's no reason we shouldn't get everything up to be solid A's. So a lot of work to do, especially on security and networking with our partners. You'll see a lot of that. Towards the other side of the canvas though, this is the platform for existing applications and new applications. And understanding new applications and where they're headed is a key part of our Cloud Foundry open platform as a service strategy. It's absolutely looking at your data-oriented architectures, understanding big data and big analytics. So we have a lot of work to really understand those communities, get involved in how they write their apps and then from a Cloud Suite standpoint, make that the best place to run them. Well, I have to kind of do a follow-up for my last question, because I want to get one more point in there. So tell us, areas, will you be doing M&A and startup acquisitions? Well, obviously we're going to continue to be aggressive on both fronts. All I can say is that the software-defined data center has many components to it, and many of them have a long way to go. So we're going to be very active in taking that exact same vision, but hopefully finding the pieces internally and externally that really help accelerate it. Okay, Steve Herrick, congratulations on the great VMworld, great to have you on theCUBE. But we'll be right back with our next guest after this short break. Thanks.