 The Cube at EMC World 2014 is brought to you by EMC. Redefine VCE, innovating the world's first converged infrastructure solution for private cloud computing. Brocade, say goodbye to the status quo and hello to brocade. Okay, welcome back everyone. This is day three of live coverage here in Las Vegas for EMC World 2014. And this is the Cube, our flagship program. We go out to the events, extract the signal from the noise. I'm John Furrier, the founder of SiliconANG, I'm joined by co-host Dave Vellante, the co-founder of wikibon.org. And our next guest is Cube alumni, Chad Sack, senior vice president, EMC Global SE. Soon you'll be president the way you're going. You keep on getting promoted every year. Chad, welcome back to the Cube. John, it's good to see you again. We love you. You never know what's going to happen on the Cube. It's your famous quote that's been played over and over again in the Tuitosphere and the social sphere. So, you know, I got to ask you one, how you're feeling, you're always dynamic and you always have a lot of energy. Area 53, area 54, is there going to be 55 soon? I mean, come on, give us the update. What's happening with you here? So my energy levels are through the roof. How could you not? There's so much awesome stuff going on. I've been talking to customers and partners for the last two days. But yeah, area 52 was the day before yesterday. And area 52 and area 53 were basically two things where we do some face-meltingly awesome, cool demos, technologies that are based in reality or in fact, in some cases, real, or on the advanced drawing boards, but near real and show people the realm of the possible, blow some minds, right? Yeah, face-melting demo has always been the phrase you've had, but one of the things that we talked to Jeremy Burton. And by the way, I mean it in a good way. Some people have said, is that a bad thing? And I'm like, no, it's awesome. You want people falling out of their chair. And as Jeremy Burton says, if you don't show customers the future, they won't stay with you in the present. I think that's one thing that you guys have done very well with the labs, you've been involved in from day one, lab work, now the future. So what are some of the things you were showing? Obviously, emerging tech is on everyone's mind and all the customers want to know as they cross that bridge to the third platform with EMC or the modern infrastructure, they want headrooms. Like they got cars, they got AC and power windows. What's the headroom that you guys are showing there? So I think there are a few really big things that we showed. And I want to say it in punchline format, so it's clear to everybody who's watching. EMC is in the commodity storage game. So we showed the Viper object store running at petabyte scale. We showed the Viper HDFS store being used for next generation applications using Pivotal. And that's something that I'd say our brand is no longer a reflection of who we are, right, where our brand is a lags the reality and frankly, the future of where we are as a company. Embracing software-only storage stacks. We did a demonstration that showed Scale.io. We have customers with thousands of Scale.io nodes. We have other customers with 50 Scale.io nodes. We're using all flash doing 15 million IOPS on commodity hardware. We showed the Viper controller making it all simple and easy. And then we showed some really awesome stuff with Extreme.io. Extreme.io is going like hotcakes. Clearly it's on like Donkey Kong in the all-flash market. We feel very strongly as a company that architecture wins and architecture is fundamental. And actually that was one of the big themes of the, you know, Area 52 was there are four distinct phylum or branches in the tree of storage life, right? And in each one of them, their fundamental characteristics define what they do. The name of the game is to have the diversity of those solutions that cover that whole swath, which we think that we do. Again, we have respect for all of our competitors, but, you know, customers are overwhelmingly voting what they're doing. So is this an inflection point? We're going to look back at 2014 and say that was the sort of beginning of a new era of storage, software-defined, software-only commodity, the beginning of the end of the box cell? Short answer? Yeah. Now beginning of end is probably not the way I'd characterize it, Dave. I think basically in the same way that all of these things have got long tails, that's one aspect. The other aspect that's very interesting is we've got customers, the hyperscale, web-scale customers, that say to us very clearly, we love Icelon, we are happy Icelon customers, but we want Icelon as software. To which we say, awesome. Done, and we actually demoed that. We've got enormous exabyte scale customers using Atmos and the Viper object stack, and they say we don't want to use your hardware, we want to just bring our own hardware. But those customers have teams of people that do iron as a service, bare metal as a service, that service and maintain a core hardware infrastructure pool. Most customers don't do that. Most customers basically say I want an appliance, even if I have the choice of software-only. And that's why I think it's really cool how we did Viper 2.0, you want software-only? Kumbaya, go to town, infinite flexibility, you can deploy it on whatever you want. But if you want it in the form of an appliance, that's why we have the EMC Elastic Cloud Storage appliance, which is the same software stack, but on EMC-provided commodity hardware. You've got to have some hardware in there. You're EMC, but you remember Halkin Maui, right? That was the early instantiation of what we're seeing today, wasn't it? And you know, I think, again, I'm very proud to be an EMC-er and I'm passionate about EMC. One thing that I think is very cool is we're not afraid to innovate organically. Maui, Halk, early days of seas of storage and object stores, we were frankly way ahead of the market in that timeframe. And that was totally organic, right? And now we have those things in the form of Atmos and the Viper Object Store at customers at exabyte scale. But we're also not afraid to do cool things in organically, right? Extreme I.O. is far and away the best all-flash array on the market and we knew that we couldn't take our existing IP stacks and make them great all-flash arrays. These guys, we saw clearly, we were way ahead of the game and everybody else and architecture wins over features. You can add features, you can't change architecture. There are architectures right. And look at this week, like how did the week start? You saw us clearly be willing to disrupt and innovate in organically with DSSD. Like how cool was that? No? No? I mean, interesting. The Rembrandt of Hardware. Oh, you're pulling on a thread. They're very interesting. The Rembrandt of Hardware was on stage, right? Yeah, I mean, Dave and I were talking last night about translating that into the ISV channel. Austin Software Developers is where, as Dave talked about, you have to throw some hardware in there, Austin EMC, but it's not, you're not a hardware company just anymore. Your software's a big part of the theme. Joe Dutty said software-driven enterprise. That's a big part of this and will DSSD enable you guys to bring the solutions, truly the market, for all flash arrays with software. I want to be really clear. DSSD is not an all flash array. I'll say it again. DSSD is not an all flash array. It is an unbelievably high performing order of magnitude, more dense, low latency pool of NAND that can be addressed as RAM and can be addressed via HDFS and key value stores directly. And I think that that's actually something that's very disruptive in the sense that you don't need to build an ISV market for that. Developers already expect those APIs as their native APIs. And if you bring to them something that basically says, I will give you orders of magnitude, better behavior. So let's talk about that pools of NAND that's addressable memory, because that's really good. That takes pressure off the spinning disk. That gives developers more capabilities. If you're, you know, we're the old model of Linux kernel, you had issues around RAM, you used disk to modify that. Now disk is now being modified with SSDs and so on and so forth. What does that enable? So in area 53, you guys did a lot of kind of futuristic things. We saw Avatar on there. I think there's some quotes I saw on Twitter that, you know, the abyss movie had 900 gigabytes of storage and, you know, 180 petabytes or something like that for Avatar. So what does that mean? And obviously the demo is how the futuristic call of duty like feel to it, kind of a gaming. But that's the modern computing environment for the customers. Soon it will be very much gamification, real time. What is the enablement? What will come from this? So the most important thing is a radical new generation of new applications. All infrastructure's job is frankly to work, be invisible, have a certain degree of resilience. Sometimes you need infrastructure resilience. Sometimes you do application resilience, but all of that stuff's job fundamentally is to support apps. You know, that's the end game. And fundamentally what you've got is you, you have the ability to create new applications in new ways when you have giant tubs of object data. And you can already see this. Like the fact, like look at you, man. You're on Twitter, you're sending stuff. Twitter, Facebook. No, we're on CrowdChat. Don't get confused. Who's LinkedIn and Facebook? Yeah, that's right. CrowdChat, CrowdChat. You got that, all right. But, you know, the point is like, it's changed the way that people communicate, the way that they share, the way they collaborate. You can't have simplicity, which is better than Dropbox, better than Box without giant object stores. That's a world changing thing. It's real time. Real time is the action right now. Would you agree? So that's at one end, big giant data tubs. At the other end of the spectrum, you've got real time analytics in memory databases, things that are enabling us to discover new insights against huge reams of data in unbelievably short time frames. And that is transactional in behavior. That's at the other end of the extreme. And developers are using that to develop whole new applications against that stack. And in the middle, you've got a new data fabric, which is things like Hawk or MongoDB or HBase, these things that basically use HDFS storage models, but for no SQL and SQL based transactional behavior. People use that data fabric to build new applications using things like Pivotal CF. And those applications change the way people communicate, learn, live, work, the way we operate as humans, right? And I think that that is really cool. The infrastructure underneath that is sexy only in the sense of how much it can enable those new data fabrics, new pass layers and new apps. So I got to ask you a question. So when you talk about DSSD, I've always loved the whole Fusion IO, atomic write piece. But we've said consistently, they've got to get ISVs to sign up. And they've got, you know, my SQL because it's the big guys haven't Microsoft Oracle SAP. What I thought was significant about the DSSD announcement was SAP saying, we are going to write to that standard. Not we are going to. We already are. Well, they were originally, they were early days, early days. But what were you, I didn't understand what you were saying about the ISV piece. What's different than say previous attempts to court ISVs? There are already well-known, well-understood, widely used, and I would say de facto standards almost at this point about how developers expect to deal with persistence. Yeah. And frankly, for a ton of stuff at one end big data tub and in the middle of those new NoSQL and SQL based apps, the expectation is lib HDFS. In other words, an HDFS library that you call that says, I am getting an HDFS object. What's underneath that? The developer doesn't give a rat's ass about it. Like, I don't mean to be rude, but they don't care. All they want is to go to lib HDFS. If we can basically interface directly in a lib HDFS, we don't need to get a ISV or a developer community to embrace our thing. Right? It happens by default. It happens by default. And there's other examples along with lib HDFS around key value stores that also have got strangely an almost object storage like semantics. And what's interesting is that it's get sort of, right? Now it's different. The get put example you're thinking about is the giant HDFS data tubs and the object data tubs, right? But those object data tubs which are widely used like things like the Viper object store, S3 Swift, Ceph's implementation would be another example. Again, we think with Viper 2.0, we have the best scale out object, distributed shared nothing object store. But what's interesting is that that object use for high performance is weird. It's something weird in you. And what's interesting is it mates with NAND. With NAND, you don't expect a SCSI stack or a file system. You expect a block of addressable persistence. And so why layer all of these old concepts on top of it? And if you look in persistence land, I can think of two technologies that do that. DSSD is one for sure, right? Doesn't have a block stack. Doesn't have a NAS stack. It has this native app stack. There's one other example. Can you guess it? Can you guess it? No. Ah, you guys, you gotta attend area 52 and see like, oh no. Extreme IO was also the only other all flash array startup that said, wait a second, why do we need to have a block layout scheme internally? Like Fusion, for instance. Like Fusion, like all of the other all flash players. Yeah, but the difference is, Fusion is doing atomic writes, but okay. What I'm saying is all other vehicles layer on this kind of concept of either a block layout scheme and maybe in some cases even a file system on top of it. Extreme IO guys basically said, hey wait a second, we can actually just look at this and go, I'm gonna ingest a 4K thing. I'm gonna do a hash against it and I'm gonna write that to the NAND, but I'm gonna de-dupe it first in line always. But I'm doing it as an object. I don't need to have a mapping scheme. I don't need to have a block layer inside that layout. I only need to present it as a block device, right? So it was interesting that basically DSSD and Extreme IO are two startups, innovative startups that basically looked at it and said, how can we radically do things differently? And again, I know I'm going on and on and on here guys, but it was a good conversation. But literally, look at that. You've got innovation in organic fashion, Viper, controller, object stack, HDFS. Innovation, inorganically, right? Which has always been something Joe's talked about. And even self-disruption, cannibalizing some of our own stuff with our all flash arrays. Disrupting ourselves with software models only and software plus commodity models too. Like, you got to love it. It's a great time to be an EMC customer. It's exciting for us. This is our fifth year in theCUBE as we were on the first one in 2010 in Boston. It's been fun to watch EMC really kind of transition the entire company. And cannibalization, you guys aren't afraid to cannibalize your own business. But you've got a big business there. I'll see you in the legacy side. So where are you cannibalizing? And specifically, are you happy with what Viper's doing? Because last year Viper was slideware. This year you got some meat on the bone with Viper 2.0. How is Viper laying in there? Because we talked yesterday about how that's elegant in the stack. So first things first. Where are we seeing self-cannibalization? The arrow of deploying high-end arrays, storage arrays, when your only requirement is low latency performance is over. Those workloads belong in all flash arrays today. Now, it's important to be very, very clear about this, right? It's good because we got Brian coming on next. The number of, I've got one of the world's biggest SaaS customers, that's an EMC customer, and they want to create a much denser footprint for how they deploy their apps. The question is, should they deploy an all flash array? Well, that logic suggests they should, but you know what? They have thousands of devices that need to be synchronously replicated from one place to another with consistency groups that are in the thousands of devices. There's no all flash array on the planet that currently can do that. Over time, who knows, five, 10 years from now. So that customer is gonna go onto a VMAX with dense flash configurations. Yay. Likewise, inside the Area 53, I asked a question, how many customers have got 500 terabytes or less? Two data centers are less. Need some block, need some NAS, need some object. Do back up to disk. Want to do some VMware, some Hyper-V. In other words, they need a Swiss Army knife. What product is that in our portfolio? It's our type one architecture. It's VNX and VNXE. Those things will continue to proliferate, but cannibalization around workloads that only were used on shared storage for low latency behavior belonging all flash, and they belong in the best all flash array, the only all flash array that does consistent low latency write behavior under all conditions, which is extreme IO. The other place where we're disrupting ourselves is with the software and caught stats with Scale.IO and the Viper Objects stack. Those are GA, customers love them, two thumbs up. The Viper Controller isn't disruptive, it's a simplifier. The Controller basically abstracts out all these different storage models because people don't want artificial complexity. They want just enough diversity in their architectures to meet their workload diversity, and then they want to simplify it. They want to pull it, abstract it, automate it, and that's what the Viper Controller does. Hey, give us a word on VVol's. Pat stood up in front of the EMC world audience and apologized for not having VVol's out. This has been a long time coming. Can you explain for the audience what's behind all that? So, VVol's is a very important project, one that we've been working on intimately for the last few years. It's true that it has taken us longer than we anticipated to GA. VMware, basically that's a two thing. VMware has to add the VVol capability to the VM kernel, the Vossa extensions to be able to communicate those behaviors to the array, and then the entire storage ecosystem in all of its forms, hyper-converge, type one, two, three, all of these different architectures have to then implement VVol's. We are, I've got a VVol breakout going on now where we're showing, demonstrating what we're gonna do. We will be their first, best, and strongest, and while Pat would kill me if I gave a specific date, we are looking good and with the firm committed date for VVol's coming soon to a theater near you. And what it means for customers is better visibility, simplicity. Simplicity, people do not want to have a data store construct that they put VMs in. They want storage, behaviors, and policy where the infrastructure's invisible and applies to the Vapp object. And that's what VVol's like. The most important question everyone wants to know is is it gonna be a Chad World reunion? Absolutely. And when and where? So, my good buddy Wade is now actually at Cisco, a great partner, and he's actually, it's awesome actually, we were talking over the last, no, not a trader, it's awesome. Yeah, we'll talk about that later in the wrap up. I think it's great. Okay, we'll see. So, he and I are working together in this new way, which I think is great. I think we're gonna do a reunion and it'll probably be in a slightly different fashion. Someone told me, Chad, you can't keep doing that shtick. You're a senior executive at EMC, and I'm like, are you kidding? I'm gonna keep doing that shtick because I'm a senior executive at EMC. Chad, I'm really excited to see you on theCUBE, great as you always, we've got to break, but I want you to cheer in your own words to the folks out there that are watching. This third platform, this EMC world, why is this point in time so different? And why is it so important? Why is this an inflection point got everyone all in a buzz, all excited with opportunity? So, building new applications that will change the world is a pretty exciting topic. Being able to do it on new paths and new infrastructures and service models that are simple, easy to consume and have public, private models, that's a beautiful thing. All of the diversity of the persistence ecosystem is fun and exciting because you get a chance to learn, play, do all sorts of cool stuff. And one last thing, talking about learning and playing and talking about self disruption, go to www.emc.com, whack, get Icelon, and whack, get Viper, get this. Those products are available for you to play with with no time bomb, no feature limits, they're fully enabled, it will never stop working for free. From EMC, can you believe that? Software only. Software only. It's great to see EMC opening up, certainly the developer community now coming down in a dev ops way to the storage layer, Viper's doing well, we'll see what happens with Demax and all these other solutions. Obviously Stream.io, DSSD is exciting. We'll see how that translates for the channel, for customers, Chad, great to see you live at EMC World in Las Vegas, this is theCUBE, as always, extracting the signal from the noise. We'll be right back with our next guest. Thank you.