 of 2012, day two, this is theCUBE, SiliconANGLE.tv's extensive coverage of VMworld. We go out to the events and we extract the signal from the noise. I'm John Furrier, the founder of SiliconANGLE.com, and I'm joined by my co-host. I'm Dave Vellante of Wikibon.org and we're here with Chad Sackich, who's the Senior Vice President of EMC, Longtime, CUBE alum, friend of theCUBE. Great to see you again. Guys, Dave, John, it's great to be here again. VMworld, I mean, you know, I have to say, so I go back a couple of years ago, two or three anyway, we were talking and you were laying out some vision. I remember the chalk talk session that we had about abstracting complexity and, you know, we're going to do to storage and networking what we did to compute it. Now you're hearing that vision, you know, really come to fruition. Was I right? Yeah, well, your direction is very clear, right? The record. So my question is really, are we, you know, we're there with storage, like, we're making a lot of good progress there. You're seeing some major action going on and networking, but yeah, I get thinking in general, you were right on. So you got to feel good about that. You really had a good grasp of where it was going. There's still a lot to do. Yeah, clearly. So, you know, the software defined data center is partially about where value moves. It's mostly about where policy is controlled, but it's also about decoupling control planes from infrastructure, right? So in storage land, we're done pretty well in terms of the policy moving up to SDRS, storage IO control, policy communication with stuff like VASA and things like that. But one thing that is still not right in storage land is the fact that the control plane for storage is always embedded in the storage target. So there's still work to do and there's lots and lots of cool stuff for future EMC worlds and VM worlds, but you can see insane progress. Well, you just wrote about VM granular storage. Yeah. It's kind of exciting, gave us a few hints there. Well, so that was top, top secret last year, right? And we were working on engineering prototypes at the time. We weren't really allowed to say much about it. And then basically I saw Duncan Epping, one of my buddies at VMware, do a blog post on it. And then VJ who did the session posted it and I'm like, okay, I guess it's wide open. Fair game. So I did the post, that will change the game in storage. And I'll tell you, it is consuming one of the largest proportion of our engineering roadmap for future years to come across everything that we're doing. Talk a little bit about why that will change the game, help people understand. So this is a bit of a weird idea. So it takes a little bit, it's like quantum physics, you kind of have to forget what you know, right? Basically today data stores, whether it's on block or NAS is a bad construct for virtualization. Because it means that the policy at the storage layer is always in a container where the container contains multiple VMs. And all of the VMs have that policy, whatever it is. If we want to get to this software defined storage world, the storage infrastructure, whatever it is, would need to be able to enforce policy on a VM by VM basis, not fake it. NAS devices can kind of fake it today by doing file level snapshots, file level replicas. But literally the storage array would need to operate in something that would be a VM or a V app level of granularity, something that's called a VVol. The other reason that it's problematic is that managing these environments at scale involves managing thousands and thousands of LUNs or file systems, which is a pain. So there's something that's called an IOD multiplexer, a name that only an engineer could love, which basically is like the IO path for everything regardless of the number of VMs. It means that in the future today, the way storage policy works for VMware is they pick what data store you place and then they move the VM to try to meet the policy. Ideally what would happen is VMware as the policy layer would say, this is the policy I need for this VM, you figure it out, and then you're the one that if you need to make a change internally to keep that policy coherent, do it. You can get to it from any available path. Right, okay. So Chad, I want to get your perspective on something because you're in the trenches, but also you got that unique ability as an executive to kind of see the big picture in kind of the 20 mile stair. Pat was just on talking about the convergent infrastructure demo HP was doing and yesterday they laid out the old way, new way, old servers to cloud, old apps to new apps with big data, PCs to mobile, et cetera, et cetera. So the question for you is I want to get your perspective on if converged infrastructure is considered old, which I believe it is, because it's been defined many, many years ago. And you now factor in that data, data, dealing with data, big data in all aspect data, moving data around, all aspects of that relative to storage, defined data infrastructure as a way to talk about it as a modern converged infrastructure. What's different about the new data infrastructure if I want to make the statement that data infrastructure is the modern version of converged? Given the market force of flash, storage, and the stuff you were mentioning. So first things first, today's definition of converged infrastructure of integrated server network and storage hardware stacks. It's really about just getting people to spend less time on the stuff that doesn't matter, right? It has nothing to do with the individual architectures of a good server or a bad server, the individual architectures of the right storage design or the wrong storage design. It's literally stop paying attention to the noise, move on to deploying something that matters, right? And you saw the HP demonstration, which I thought was cool, but hey, we did UIM demos a couple years back and so on and so forth. What is changing is these new requirements at the storage layer are coming in a couple of different forms. The first form is when customers need Swiss Army knives because they're small and that's the only infrastructure they're gonna have for storing all their data. They're small data, they're big data, they're NAS, they're blocked, they're object. Architectures that are simple, easy to consume, buy, transact, go through the channel, those are the things that are winning today and they're gonna win tomorrow. Because of the packaging and simplicity. Packaging simplicity and their general purpose, right? And I think you can expect to see continued growth of things that look like hybrid arrays, that have Swiss Army knife-like functions and VNX as EMC's answer in that market, right? And VNXE, but what we're finding is as soon as a customer has the requirements that says, I'm okay with not Swiss Army knife, right? I want a design for purpose. Immediately they diverge into three different architectural models that are just totally different and you can take the EMC product- Where data is critical. Right, and it's a sufficient scale to warrant this. They need it for transactional systems where all of a sudden the all-flash world starts to become very, very important, right? And what we saw is we saw that the requirements, by the way, in all three of these categories is that scale-out is a fundamental design construct. So basically in transactional systems, scale-out models that increasingly, and step-by-step-by-step will move towards all-flash models just intrinsically based on technology will come to dominate. And we looked at it and we saw that Extreme I.O. was, in our opinion, the best technology out there, right? Then you could buy. That's right, or that existed, right? So we're doing demos here that show hundreds of thousands of IOPS all doing in-line D-duke, all things that you just could never do if you didn't build it in. Okay, so let's talk about flash, because I want to do traffic, go ahead. Very briefly, John, right? So that's category one. The second category is the scale-out NAS. The fastest growing category of storage infrastructure that's in this space, and you can see us and others battling it out furiously in that space, right? Icelon is growing faster than we can manage it. Like our biggest problem is how do we grow our resource, our channel and all that stuff. But the third category that is also scale-out in its nature is basically analytics storage, right? Whether it's unstructured with Hadoop or whether it's structured, massively parallel DAZ systems, right? So each one of those is solving a very unique and different architectural problem, but they've got these fundamental characteristics of scale-out, built-on commodity parts, software-based, and their characteristics are suit to need. I totally agree with you. And I want to ask, and it's causing a lot of confusion. First of all, those are legit use cases and architectural constructs. Now the question now, because we're coming back down to the old days of interoperability conversations, integration support. So the question is that it seems to be, everyone's definition seems to be different because their architecture is different, because of their environment. Are you seeing any new, is that true? Do you see this data infrastructure? And how does that make data infrastructure different than just saying, hey, I'm buying some conversion infrastructure? What's interesting is these architectural models flow from their use cases. Check, agree, agree, agree, right? But what is unfortunate is that there's not a decoupled control plane. Wouldn't it be nice if there was some sort of like hint, hint, abstraction layer that would not just be for EMC stuff, and not just for these three categories of transactional blocks, scale-out NAS, and by the way, the transactional block is scale-out in nature. And the scale-out NAS use cases and even these object storage models, wouldn't it be cool? There really needs to be an open flow of storage in a sense. Some sort of software solution policy-based homogeneous infrastructure. Put a top-down data management strategy. But the thing is that it has to not mess with the data path. It needs to be control path only. So I'll just jump in and throw a wrench into the equation here. So HP ships an open flow hardware device in their quote, converged infrastructure. We have Bethany Mayer who leads that networking group, and they have a conversion infrastructure. They're organized beautifully around the old convergence. But yet no one talks about open flow on HP. I think they're the only ones shipping a product. Is that because it's not relevant in the architecture? So I think, as I was writing the blog post from last night, I was thinking to myself, it's funny, people will look at everything and analyze it. And analyze it. And like for example, VMware's doing some tech previews of things called vSAN, EMC's doing previews of things that cannibalize our other thing. And everyone always wants to jump to the conclusion of like, okay, who's eating who, and who's doing what, and it's fun drama to watch. And it's fun, it's exciting. It's IT, right? Yeah, yeah. But what does a relevance equation, though? Well, what people miss is the biggest hazard for any company isn't that you get eaten by somebody else. It's that somehow you lose your mojo and your willingness to cannibalize yourself and innovate, et cetera, et cetera. And the challenge usually is related to your business. If your existing business revenues are over here, it's very difficult to go and say, I'm going to eat myself with this thing over here. One thing that is very interesting is as we do this, there's value that moves all over the place. We launched this VDP thing, right? Which will no doubt eat some of the low end of Avamar's business, but that's okay. And it's all good, right? I think that in the case of the HP thing, they need to make that team literally have a charter that says your job is to kill the old way. Yeah, not just a feature. It's not a feature. Your job is to kill the old way and set up some... Well, successful companies eat their own before the competition does. That's clearly an Oracle decided to take the different approach. They just kind of protect this whole nother. Now there's no story, but... I think it's, Joe said it. Joe's a man who's just a fountain of genius quotes. Tucci, right? He basically said the single defining characteristic of all the companies that die in each one of these technology waves is they're playing a defensive game. Right, so I'm not going to tell Larry Ellison what to do. He owns an island. I do not. So clearly he's doing something right. He's like a telco. Protect your territory, extract rents out of the ecosystem. You know, there's probably people inside Oracle going these in memory database models, these NoSQL database models, Cassandra, all this stuff. We've got to get in there. And someone within Oracle, hopefully for their sake, is saying, yes, we will protect you from the natural organizational tendency to... Or maybe they're saying Hadoopa snake oil. Yeah, right, you know what I mean? Yeah. Well, what's the big disruptor that you're seeing right now? I know time's getting tight here, but I want to ask a couple of questions around just observational chats. So you don't have to, no one will take read into it other than just your genius. But share with us your observations of what you see disrupting out there. What's, you know, just the Chad view of, you know, what points of disruption you've seen some smoke and where you see fire going on today in the market. Biggest disruptors I see over and over again are totally new data models, totally new application models. Far out in a way, like, look, there's disruption in the world of flash and storage enormous, right? There's disruption with SDN and networking enormous, right? There's disruption of open stack entering into the market and gaining steam, right? And you can see VMware responding this week, in my opinion, you know, looking at VMware from an outsider, responding to changes that they're seeing in the market, embracing that. Dynamic ops, the Sierra and others, right? But far and away, when I talk to customers, the thing that really basically changes their business is when they have new data models and that's partially like database models, it's partially big data and analytics. And as they entirely build new apps using new app frameworks, that changes the game entirely. In some cases, it doesn't necessitate traditional storage, virtualization at all. So my question next one is more organizational around the industry, because you had the experience in following your career of doing a lot of integration work with VMware as part of the EMC team. So you've seen under the hood, you worked with the players, you were in the ecosystem, not only personality, but also as a technical person and executive. So you've worked with the VMware machinery. Yep. So okay, here's the premise. VMware's growing like a weed. They're expanding beyond the VMware focus of just VMware. I would say more like a flower. Like a flower, more like a beanstalk. So now they're going multi-vendor, they're embracing obviously the partnerships. And we're seeing their moves like Nacir and some of the other things that are just, like Joe said, it's a rapid disruption. What's your advice to people like VCE, like NetApp, other vendors and other startups that have to kind of make quick decisions around how to adjust to VMware? Is there any advice, observation you can share with them? Because I'll see, when you make moves like Nacir, it changes the game on the whole architecture because now you have an emerging software component, policy based, blah, blah, blah, and VCE is more of a high end thing. VCE in particular has to make some, just some tweak their positioning, change their messaging and do those kinds of things. So that's a very good question. I think that the reality of it out there, right, is that every technology company is looking out for their customers, if you don't look out for your customers, something's going to go horribly wrong. They're looking out for their shareholders and they're looking out for themselves, that those are all like one little idea package, right? Yeah, they make their moves. And they need to, each company needs to think about what is in my strategic short-term, long-term interests. And that's just a fact. I'm not saying anything that's earth-shattering there, right? I will say that VMware has got a track record of creating shifts in the industry that represent, to some degree, terrifying changes to the old, but turned out time and time again to represent huge opportunity, huge triggers of innovation. But they're not surprising anyone. If you're in the know, it's pretty obvious, they know what the maps are. The trick though is that when you first find out about something that you look at and you go, wow, this is going to have some impact. If they're successful with X, it's going to impact something Y and that's going to cause some disruption to my little company ball of customer, shareholder, self-interest, right? You can either curl up and fight and that could be a winning strategy. It hasn't proven to be a winning strategy to date, right? All of us have to basically out-innovate. And don't get me wrong, VMware's got companies that are trying to out-innovate. So basically, to summarize, you put your running shoes on and keep up. Totally, man. Basically. Totally. Nail your value proposition, make your tweaks you need to do and keep your running shoes on or be left behind. And accept that basically the only constant has changed. The things that were your core value proposition at some point may not be your core value proposition tomorrow. So today, there's a set of value proposition that leads more customers to choose EMC for VMware than anyone else. Great. How will those change over the next two or three years? They will change. How has the community changed? Here's a question for you because you can get some insight on this one. How has the community of VMware, VMware ecosystem changed over the past 24 months, 12 months, and how do you see it evolving? Obviously it's growing, so that's one factor. New new entrants, new migration of people coming in. So how would you see what's changed and where, what do you think it's going to be going? Someone actually asked me this online. They said, now that VMware is the 800 pound gorilla, they're no longer cute and cuddly and fight the man and they are the man. How'd that happen? Stick it to EMC. So the thing that I responded with is the community around VMware. And here I'm not just talking about the vendor and partner community. I mean the community community. People, the people, the real people. They gravitate towards VMware to Vmugs to VMworld because it's really cool and great technology because it's helping them with their business and because the community is a self-fulfilling thing. So long as VMware can keep cool technology that's helping people with their business, the third one of a community that is self-fulfilling will keep happening. I mean Vmugs are bigger now and more interactive and more intimate now than I've seen them before. Yeah, they're on governance in a way. People do stupid stuff at VMworld like those reserve for Chad Sack Edge things. And BJ Jenkins, you know, I'm gonna come and I'm gonna get you, man. I'm gonna get you, BJ, right there. But you know, this place is fun. I mean like, I don't know whether you guys feel the same way, but this is one of the most fun dynamic communities. I don't think that that is changing. I think it's solidifying. What would you say to the VMware employees that now inherit the greatness of Pat Gelsinger who should do a great job. He's young, he's solid, he's got great experience, great knowledge, great technical knowledge. What would you share with the folks that watch us from that are employees of VMware and they're partners? I've had the great luxury of working with Pat from the first day that he came to EMC. A funny story, I don't know how much time I've got, but I'll try and make it short because this is a gas. You know, Joe actually goes, you know, tap, tap, Chad. Pat's coming in, give him the brain dump on everything EMC. And I'm like, okay, great. So I sit down, I'm on the whiteboard, I'm explaining it. And then he started to ask some questions that illustrate he had a fundamental lack of knowledge about the storage industry as a whole, right? And I'm like, wow, this is going to be funky, right? Literally a week later, I'm back and he's arguing about the code path inside ingenuity with Symmetrix Engineering. So in the time span of one week, right? He hit the books. Boom. That's a quick study, keep up with us on theCUBE. That's one thing we know. He had that Matrix program. He's an engineer's engineer. He accepts nothing other than being the best. He's very passionate about that. He's very focused. And what I would say is, I think it's one of the best examples that I can think of in recent history about exec transitions, right? So what VMware needed when Paul came in was a vision of where VMware could take the entire industry. And if you look back, what he said way back is basically still kind of directionally where we're going. It's gone through different names, cloud OS, data center OS, software mainframe, software defined data center. It's always been this idea of abstract pool resources, make them elastic, which allows people to focus on the stuff that's up higher, right? And then over the years that included a lot of different strategy changes to expand into the app. The vFabric data director thing was a joint EMC and VMware project. Lots and lots of stuff like that, right? What's needed now is that focus on execution against vision. There's nobody better in the world to do that than Pat Gelsinger. Okay, we have to cut it short with Chad. Small interview turns into a long, great content machine. You are a content machine, whether you're doing Chad's world, jumping out of airplanes, or just being a great techie and a great executive. Great to have you on theCUBE, great to see you. John, it's great to be here. We'll be right back with Mark Egan from VMware. We're going to talk about what's going on at the CIO level, how these guys are rolling out and how they eat their own technology. They live it, they're use casing it, so we want to hear cutting in stuff from VMware inside the ropes at VMware. We'll be right back with Mark right after this short break.