 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. We're here live in Las Vegas for EMC World 2014. This is the Cube, our flagship program. We go out to the events, we check the signal for the noise. I'm John Furrier, the founder of SiliconANGLE. I'm Joe, my co-host Dave Vellante, co-founder of Wikibon.org. We have a special spotlight drill down session with Extreme I.O., Josh Goldstein, VP of Product Marketing and Management of EMC Extreme I.O. Welcome to the Cube. Yeah, thanks John, good to be here. We love doing these drill down spotlight segments because we really want to get into the weeds on kind of what the deal is with Extreme I.O. Bombs are dropping everywhere. Tucci, Goulden up on stage, talking about the acquisitions of DSSD and all the action going on in Flash, third platform. You guys have been in the center of the action for quite some time now. So the horse that was been riding on the EMC Flash strategy, give us the update, what's going on? Give us the quick highlights that we can jump into some of the stuff going on at the show. Oh yeah, sure. So I mean, if you remember, this is, at EMC World here, marks the two year anniversary of when we first were acquired and first showed Extreme I.O. technology to the world. And a lot has happened in those two years. Today, you probably saw the big announcements about having a $1 million guarantee on Extreme I.O. And this is in line all the time, meaning that we have in line data services, they don't switch off, they don't throttle back, we don't have system level garbage collection, all the things that are plaguing other flash arrays that have been in the market up until now on Extreme I.O. These are just not issues you ever need to deal with. And for a customer that means it's the most consistent and predictable performance that you can get. And a million dollar guarantee, that's also if someone can prove something and also you have a rescue program, right? We do. I saw the rescue program. The flash rescue program, right? Yeah, and what we see is we got lots of customers who have invested in another platform before Extreme I.O. came into the market. They've been in production for a little while now and they're starting to see some of these weird effects of the array not performing the way they would want it and we'll make it very attractive for them to switch to Extreme I.O. and never have those problems again. Okay, but so I want to ask about the million dollar guarantee. I mean, you guys provide in line data services, right? That's right. So this guarantee is like kind of a no brainer for you. You're not going to pay out on this thing. That's the whole idea, right? Okay, so it's a good way to get people's attention. It's a good way to get to pass lawyers and things like that. If you have a guarantee where your product will never fail in that manner, then you have no issue. Okay, why is it so important to have in line? Let's start there. Well, it's important for consistency and predictability. So when you're buying a flash array, you're typically investing in a platform that you expect to have higher levels of performance, lower levels of latency and to be very consistent and predictable. That's the thing that flash should offer you above and beyond a disk or a hybrid system. And when you have data services that are post-processed, you can't guarantee that kind of consistency and predictability anymore. You never know when that post-processing operation is going to kick in. You might decide that you're going to try to schedule it in the middle of the night, but in today's global environment, there is no middle of the night anymore. Things are going all the time. So the arrays can get into situations where they've had to throttle back the data services or switch them off. The capacity on the array could start to balloon because it's no longer deduplicating. The latency can get higher. Sometimes worse than disks. Sometimes it's worse than floppy disk, which is hilarious, but we see it over and over and over again in these real environments. Now, okay, so now this notion of in line, it's not exclusive to EMC, but it's certainly not broad based in the industry. Can you talk about that a little bit? Well, yeah, so let me define in line and what we mean by that because a lot of people use the word, but really not very many people are following it the way we would say. When we say in line, we mean that a duplicate write is never hitting the flash in the first place. It's not even landing there for a nanosecond and then being post-processed later. So we are finding everything in line in the data path and eliminating it while the IO is still in the memory of the array. It's never being committed to the flash in the first place. Okay, so I wonder if we could talk about the data services for a moment. So what data service are we talking about? You mentioned duplication. What other data services are you talking about? So talking about things like thin provisioning. One of the things that we're bringing out in our release that comes out in a couple of weeks is encryption. That'll be 100% in line. All of our snapshot capabilities are in memory and in line. This is another thing that is new for us here at the show. So tremendous value in being able to do these things as in-memory data services in terms of speed, predictability, performance, and even new capabilities that you couldn't do in the past. Josh, I want to ask you about your stack inside Xtreme IO. Where's the stack come from? Is it all built from scratch? Are you borrowing certain things from EMC Classic? Talk about that a little bit. Everything at Xtreme IO is purpose built and was built by the Xtreme IO team. So this is really truly a ground up all flash array architecture. And it's not just that the stack comes from Xtreme IO, but even the concepts of how the array is architected and how it operates were all built from day one by Xtreme IO in a very unique fashion. So we don't borrow things like classic things that you would see in disk arrays. We don't do things like log structured file systems. We don't do things like raid algorithms that you would have found in disk arrays. Everything was built around the idea of having a random access media and knowing that we were going to have that from day one. So built for flash. So that's good news because modern, at the same time, it's hard to build a stack. Yeah. So you've been at it now more than two years. So talk about how the stack is evolved and in particular how it's evolved and subsequent to being acquired. Well, so what's interesting about the way we built the stack is we had a very strong idea of what we wanted to include in it and what the architecture needed to look like from day one. So there's things that have been architected that still are not released features, but they've been envisioned from day one so that we would have the ability to roll them out later and still have them not be bolt-ons but really be integrated as part of what we're doing. So for example, snapshots, which is the thing that we're showing here at the EMC world, that has been in the code since day one, but we wanted to, when you talk about what's changed as being part of EMC, one thing that changes is the bar to release enterprise storage array is very high. As the leader in storage, we have to hit a very high quality bar and we want to make sure that we're testing under very strenuous, very realistic and very long-term scenarios. So that means that we're going to really soak things in our QA. So snapshots is a good example of that. We could not just have a feature and say snapshots are out, but we want to do big permutations of snapshot trees, take snapshots of snapshots of snapshots to many levels deep in a snapshot hierarchy, do that on many volumes, keep changing the snapshots because they're writable so allow them to age and make sure that this is something that is really robust when we release it. So you guys actually, I think coined the saying, there was Jeremy Burton on theCUBE at Oracle OpenWorld, said a little bit of flash goes a long way and our counter was, imagine what a lot of flash does. At the same time, you saw David Goulden today talk about some IDC data. By 2017, I think only a small portion, maybe 12% of the capacity would be flash, but of course he wasn't talking about the dollars, right? The dollars are going to be significantly higher than that. And you know, we've been very forceful about our position on all flash. So talk about some of the workloads that you see. Obviously a lot of database applications. I want to ask you about that and I want to ask you about the impact on DBA productivity. Oh, it's tremendous. So I'll give you a great example. So the classic way that people would think about a flash array in a database environment is to solve a performance problem around the production database. And that's really the state of what's been done in the flash array market for, you know, it's maybe since 2009, 2010. And ExtremeIO can certainly do that and do it very well. The power of what we can do though is in consolidating the entire application environment that surrounds a production database. So in addition to production, people have development and test copies of their databases and those haven't been good candidates for flash because it's been too expensive to deploy a flash array for those kinds of instances. Then on the other side of the production database, you might have reporting and analytics applications. Again, flash has only been either not deployed there because of cost or very selectively deployed only to the hottest, most important parts of those workloads. With ExtremeIO and with our snapshot capability that we're bringing out today, we can do things that have never been possible before not just on another flash array but on any storage array up until now. We can take that entire environment and serve it off of one instance of the database. So what I mean is you can have a production database. You can take multiple snapshots to serve your development test copies. You can take multiple snapshots to serve your reporting and analytics copies. And this is all space efficient from a data perspective. It's space efficient from a metadata perspective. Every snapshot is writable if you want to make changes to them. For example, for development purposes. And they all perform just like the production database. And because we're a scale out array, we have the IO density to support huge application development environments and to give everybody high performance copies but have the footprint be very small in flash. So we're changing the economics of how you can do these things, making it easier for the storage admins, making it easier for the DBAs and really changing the productivity of the entire environment. Well, as you well know, let's call it a little bit of tension between the DBA and the storage admin. Storage admin sometimes doesn't necessarily give the DBA the visibility that he or she wants. How is flash changing that? And are you seeing any examples of that in the field? Well, I don't know if I would say that flash is changing that, but Xtreme IO is certainly changing that. But we do see examples of that. So in our cases, the storage admin is now becoming the DBA's best friend because he's the one who's coming in and saying, how can I help your application run better? Not just the production instance of it, but let's go and look at your development test environment. Let's look at some of these underserved areas because I can give you better performance and I can do it cost effectively and that's not something you could ever do before. I want to ask you about market leadership. You guys have said that you're number one in flash, some huge number of exabytes shipped. Now, of course that includes all the flash that you're stuffing into existing Vmaxes and so forth. But then I think it was either Tucci or Goulden said, oh, by the way, we're also number one in the all-flash array business. That's right. So I wonder if you could, might be somewhat surprising to some folks, although maybe not. I mean, this is, I guess, what part of being acquired by EMC does for you, right? But talk about that a little bit. How are you able to achieve sort of that market leadership so fast? We have an outstanding sales force and we have a great product that's kind of a deadly combination out there. So we spent a tremendous amount of effort and people would say to us, after we got acquired, why did it take so long for Xtreme IO to go GA? Well, there's an awful lot of stuff that you need to do at EMC scale to make this product globally available, supportable, have everybody trained up and ready to go when we turn it on day one and say it's GA. A lot of little things, right? A lot of little things. And some really big things. And some big things too. And so when we went GA back in mid-November last year, what we were able to achieve in the last six weeks of the quarter from that level of preparation ahead of time was tremendous. And so it catapulted us very quickly into the number one position in the market in just half a quarter. So I got to ask you, sort of you and I have known each other for a while, came from a smaller company, you did a startup and now you're in a big company. Any big surprises coming to EMC? What shocked you the most? Oh, I mean, well, so when it's a lot of new names to remember, I'm not used to working in a company with 60,000 employees. So that's a lot of people to get to know. But honestly, the best thing for us has been that we have huge resources that we can leverage and that the company has been so good and willing to invest in Extreme I.O. To make this product successful and to realize that it's so important to the future of not just our product line, but to the company. And EMC is not afraid of Extreme I.O. being successful. I know that people will kind of posit, you know, EMC might be worried about Extreme I.O. cannibalizing other product lines and I can tell you that it's really not the case. EMC wants to win the deal and to be a good provider to our customers, whichever platform that happens to be. Yeah, I mean, I buy that. I think the bigger, you know, question that a lot of people might have is, okay, you know, EMC made a call early on in the Flash game. And then they bring up the example of Avamar. Was this the next Avamar? In other words, they bought Avamar really early and then had to go out and buy Data Domain, which ultimately ended up being the leader in line, by the way. What gives you confidence that EMC has made the right call this time? Oh, well, I mean, I get a lot of confidence in it. I believe 100% in our product or I wouldn't have been with Extreme I.O. in the first place. And I think that EMC sees the need, you know, we're a portfolio company. So there's not one disc product. There's not one hybrid product that can service the broadest, you know, the whole broad storage market. It's such a big market. There's so many different varying requirements that there's no one thing that's going to do it all. So, you know, you saw the announcement about DSSD this morning. Yeah, I was going to ask you about that. And so this is just indicative of that. There's going to be overlap in things and there's segments for various products. Extreme I.O. is one of them. It has a very big market that we can address, but we can't address 100% of everything. And so we're going to keep investing in other areas, DSSD being one of them. You know, John Tucci says overlap is better than big gaps. And he's right about that. You can throw it out of overlap, Josh. So I got to ask the question people want to know is, okay, so what's changing? What is going to be the new impact with DSSD? Obviously, flash is not a groove. Swing at this point certainly is a straight and narrow in terms of marketplace. You know there's a customer base, you know there's a need for it. But the products roadmap still is, there's still some stuff that might move in terms of some variables in the marketplace. What are those key areas that you're watching on the roadmap where there's an obvious straight and narrow, no curves on the road, just pedal to the metal, full throttle, and some areas where you got to slow down a little bit, kind of watch the market. Oh, I don't know if there's any place where we need to slow down and watch the market, right? We have a very clear idea of what we want to build into Extreme I.O. So pedal to the metal. It's pedal to the metal. You're saying straight and narrow, pedal to the metal. Yeah, and it's not just me saying it. We have a lot of customers now and they have things that they want to see in the product. We have a very large sales force that is out there representing this product and they have things that they want to see for their customer. So our challenge is just getting that in there as fast as we can. I guess let me see if I can translate that. So basically you went to direct availability, which is basically like everyone wants the product, right? Demand was high, pretty much that's well known. Now that the demand's high, you have to deliver that. That's what you're saying. The product manager, what's on the roadmap that's off the straight and narrow that you guys are watching very closely. Oh man, I don't know how much I want to go into things that we're looking at on the roadmap that might tip our hand a little bit too much, but there's very interesting things we see in terms of levels of scale that we can do in Extreme I.O. cluster. So you know today we're selling in clusters that go up to four EXPRIC building blocks, but we're absolutely seeing demand to go much larger than that. So that's one area that you'll continue to see improved scalability. There's many data services that we still want to put into the array. So you're going to see things coming from us. I'm not going to talk specific timelines, but obviously Snapshots is in now in the release of- So inline all the time is all your key message right now? Yeah, absolutely. And it will continue to be. So there's more things coming that will also be inline all the time. Inline all the time. Well thanks for coming on theCUBE. We really had a great drill down there. G.M.I.O., pumping on all cylinders, pedal to the metal, cranking it out. Thanks for coming on theCUBE, really appreciate it. We'll be right back with our next guest after this short break. All right.