 Welcome back. This is theCUBE. We are here at HP Discover 2013. We're in Las Vegas. This is HP Discover's flagship program that their customers, their partners, the tech geeks and gurus, and here really sharing all the new announcements. This is SiliconANGLE's theCUBE. We go here where the signal is, and that's where we are. I'm John Furrier, the founder of SiliconANGLE. I'm joined by co-hosts. I'm Dave Vellante at Wikibon.org. Craig Nunes is here. He's the Vice President of Marketing for HP Storage. Craig, welcome back. Multi-time CUBE-hardened guest. Thanks for coming on. CUBE-hardened, yes. Thank you for having us here. You guys, this is a great set that you guys have. Awesome, HP. How many people are here this year? Gosh, I'd be just guessing, but it's probably around 10,000. Okay, so that's... It feels great. I mean, it feels big, but it's comfortable. Yeah, it's comfortable. Yeah, no, it's great. So you guys got some hot news. We got the deep dive. I said to Milan, we got the Kool-Aid injection a couple of weeks ago, which was good. We had a really... I liked what you guys did. You brought in a lot of people and took us through this sort of road map, and so we know a lot about what you announced today, but why don't we share with the audience what you announced here at the show? Sure, yeah. There were really a couple of things we talked about. One was in the area of three-part store-serve and an all-flash array, the store-serve 7450. And the big deal here is this is a platform that brings the performance, the efficiency, and the media endurance characteristics that you got to have with flash. It's why people are buying them to accelerate their applications as cost-effectively as possible. So you got all that, but what we also brought was the system resiliency and availability features that tier one, you know, mission-critical data require. Also the, you know, live data mobility through Federation capability to get stuff on and off your all-flash array. So it's the best of both worlds because we have customers saying, look, I'm concerned about traditional architectures. They can't address the performance and efficiency needs that I want to use for flash. There's a lot of new companies out there, paint still wet on their architecture and services. So I'm kind of nervous about that, and by the way, not a lot in the area of resiliency. So, you know, we think we've kind of got the right combination for what folks are looking for. Yeah, Federation is something we didn't talk to Milan Shetty about. That's very unique in the industry. Donna Telly this morning sort of set it up and said, look, you either have a disk-based controller that, you know, can't handle the flash, that's, you know, trade-off one, or you have, you know, modern, you know, Silicon Valley startup flash with no stack. And so you guys have come up with the best of both worlds. I call it the God Box flash. And so now I got to ask you, so you like made your mark with three-par of attacking companies trying to bolt on thin provisioning and did some damage, you know, in the marketplace, doing more damage today, significantly more than you did when you were three-par now that you were with HP. But so a lot of people are going to say, hey, that's a bolt-on. Why is this not a bolt-on? We talked to Milan about it a little bit, but give us your marketing take. Yeah, so I would expect folks who don't have products or capable products to sling that kind of stuff. The reality is, look, don't take it from a marketing guy, lift up the hood and take a look at the technology. And, you know, what we introduced today was not just a three-par array with SSDs in it. In fact, that was kind of the end of the project. The beginning of the project was all about some really important cache optimization features we drove. And when we drive some granularity in the I.O. between the controller and the flash that you can only find in a handful of vendors in the industry, I mean, we'll do 512-byte reads, 2,000 times smaller than a BNX, for example. That's important because latency is important on the back end. So we've driven some important optimizations in the software. We also had the foresight to inject an enhancement into the three-par ASIC to give us a level of silicon-enabled acceleration. We've taken advantage of that ASIC in the HDD world, but it was a great opportunity to really use that here to drive some of the performance we're looking at. Well, like I said, let's not take the marketing guy's word for it. The market's going to decide, ultimately. And so you and I have had discussions over the last several years about sort of rethinking the definition of tier one storage. Three-par, obviously, is really the only new entrance into that whole tier one space. You know, you've got the high-end IBM and EMC and Tachi and the three-par is right there, clearly stealing share in that high-end IDC price band. And then all of a sudden flash comes along, this notion of all-flesh arrays, and one can say, wow, these are the new tier ones. You call them tier zero, but you can see them if they have the stack, you know, replacing a lot of that value. And who knows what happens to that traditional tier one. And so the market's going to decide, you know, whether or not your architecture is worthy of doing that. And you guys have made that bet. And, you know, so far it's held up the test of a decade. I would also tell you, we are where we are today. And we're able to do those enhancements today because of some very smart guys, you know, a few years ago, right, the guys who put together the architecture, the three-par operating system. That, you know, that was a labor of, you know, trying to unlock the back-end performance of an array, right? Back-end performance is limited to spending media, a couple of hundred IOPS per disk. And, you know, that's kind of how that world revolved. And the whole approach they took was, how do I aggregate every drive in a thousand-drive array and, you know, hammer that through the controller infrastructure? So guess what? Dual controllers? No way. That isn't the approach they were taking. It had to be a very clustered approach where they could aim all that frame of horsepower at that workload. And guess what? All we're doing is replacing a thousand drives with, you know, 50 flash drives, right? It's not a big change for the system. The real change came in that cash-to-flash optimization. So, you know, I want folks to understand, you know, we were, you know, we were lucky, we were good, you know, whichever way you want to go on that. We have that architecture that, frankly, you know, folks who are starting today to do this are taking a page from our book. Thank God for 3-par because HP's story should be deep-duty without it. I have to say that's because you got, it's true, you got, and this is what's so interesting about your business. I mean, you got the huge install base declining and you got 3-par growing like knots, you know, 80% plus growth. So you got to throw even more gas on that fire because you got to get the new stuff outpacing, you know, the old stuff. It looks like you're almost there, you know, really, really close. I've said a number of times HP in general across the board might have to shrink to grow. I think it's true for, you know, a lot of the divisions within the company. So, to me, the all-flasher race throws gas on that fire. Yeah, I, uh, first of all, I have to tell you, the real secret to 3-par success, absolutely killer marketing. Well, yeah, well, of course. It's like the products don't really, you know. The products are table stakes. So the thing about this all-flash array, so we're talking to a lot of folks here at Discover. I mean, it is packed around the booth, you can't even get close to the system. And a lot of folks, I mean, it's hot and it's sexy, right? For a lot of these folks, they're, they might not buy one. Why? If you don't have the right application to, you know, drive the upside for the business, you know, flash is going to, you know, cost you a few bucks. Guess what? They're going to turn around and buy, you know, a 3-par store serve 7,000, the mid-range array we introduced in December, or maybe they'll go to a 10,000, a high-end array, and they'll run some combination of HDB and SSD. But the, you know, the lights go on because they're going, wait a minute. You started with tier one high-end, you know, for the cloud. You went to the mid-range, you know, $20,000, $25,000, you know, entry price point for the lineup. Now you're doing flash. This platform really can do it all, right? And so maybe I don't need flash today, but I even want to buy into a platform that can go wherever my business needs to take it, right? So that's been a huge takeaway. And for the guys who are taking advantage of 3-par today, wheeling in an all-flash tier, they're going, man, I'm there, because I already get it. And I already am going to see how I can take advantage of services across those platforms now. So that's part of my architecture. What about the economics of flash? Everybody, you know, two years ago, was like, flash is too expensive, flash is too expensive. But, you know, you think about the tier one arrays, they got 15K RPM drives, they're short-stroke of them. I mean, that looks like a dead duck. Yeah. And so can you talk a little bit about the economics? And, you know, cost is an issue, right? I mean, it's never, I mean, all things being equal, you know, going to come down the cost, and everything's not always equal. But, you know, markets, again, markets going to decide whether or not you guys, you know, are going to have the most cost-effective platform out there. You're betting you are. But talk about the economics of flash. Yeah. So, look, I never take a bet that I don't know I'll win, right? If I am going into... You've been a lot with, like, 10-year-olds. Yeah. I would never take advantage. Well, maybe I would for certain friends of mine. But the compaction story around flash is a huge enabler, right? Because otherwise, you know, the 10-to-1 difference, or whatever it is, is going to hold folks back from adoption. So that the... I pulled some stats from our installed base, right? And just looking at the last quarter, I pulled a sample that said, looking at all of our thin volumes, exported to host, you know, what the server saw in capacity, the actual provision storage, one-third. One-third of what hosts thought they had, right? Real life today with customers. And we know that, you know, thin technologies, the zero detection in silicon, we know that runs as fast as you can pump data through it at flash speeds, right? So how about... Here's the value proposition. How about you buy one-third the capacity and still get all of the flash performance you want? That starts to bring flash into, you know, into a lot of folks' budgets. And so that's why we think we've got it. Have you ever had to pay one of those guarantees? Never. Never paid it? Not once. No, so you're mentioning that. And I have to say, we know that because we don't just take your word for it. We actually analyze the metadata coming out of customers, controllers, three-part controllers, we did a statistical analysis on it, everything else, and the three-part nailed it. So, okay, so that's cool. You got some other knobs that you can turn over time, whatever, compression and dedupe and so forth. So that's cool. All right, what else? We're sort of running up against the clock here. I can't stop talking about all flash arrays and three-parts. The other thing that I got to tell you, the other thing I got to tell you about is, you know, our big push in software-defined storage. And this is something that, you know, the words are, I think, pretty new, right? We're dealing with software-defined data centers, software-defined storage. Yeah, less than a year old. Yeah, but would you believe that we've been doing software-defined storage since before it was cool, since 2007 with our store virtual, virtual storage appliance, formerly known as Left Hand. And we've seen, you know, a nice uptick in demand since a lot of the discussion here. And we brought that notion of, you know, a full suite of storage software provisioned as a VM to information protection. And so our store once line up, our purpose-built backup appliance running deduplication from HP Labs. We took that software from the appliance, and we now are delivering that as a virtual storage appliance, a VM-based backup appliance. So great for, you know, remote office deployments where you want to keep a local backup or an SMB who just didn't want another box. It's the perfect convergence approach for them or service providers who are looking to spin up a very flexible infrastructure for their customers, spin it up, tear it down, you know, SP is an interesting, absolutely very interesting. Do you have an API? Can I get a tunnel in through an API to that platform yet? Well, as it turns out, the store virtual VSA is manageable by OpenStack, manageable by Cinder. In fact, we've had a lot to do with the Cinder work. I was on GitHub the other day confirming that everybody's talking the game. It's HP, IBM, Red Hat, SolidFire. Yeah, it's a big contributor. AMC's in there a little bit. Yeah, yeah. A couple others, maybe I'm Citrix. Yeah, yeah. It's, we love the model. I mean, we've seen how it's gone on mobile phones, right? Android's really, you know, become the dominant operating system out there. Open source is a great way. Yeah, so SDN, you're there and you'll evolve that. This idea of storage as a platform is pretty interesting, you know. Yeah, yeah. But stack is the orchestration layer, API is in. Yeah, yeah, leverage your infrastructure. And I would defy you to figure how much CPU resource and how much disk is unused in these servers these days and just think, hey, if a little software can enable, you know, shared storage can take advantage of older assets and, you know, show a decent ROI, why wouldn't you? Well, you've got, you know, the architecture of 3PAR, you've got the virtualized, you know, back-end and abstracted, you know, the physical hardware. Absolutely. That's good, yeah. Heterogeneous, that's an interesting play. That's something that you guys have never been big on, you know, connecting up heterogeneous devices. Well, in a way, you know, what we're talking about here and the beauty of a VM-based storage device, the CERT matrix, the QA matrix, is no longer physical from our perspective, it's the hypervisor. Yeah, okay, so it's up to. So, it's up in stack, VMware. It's up to, yeah, VMware and Microsoft and, yeah. And so, we tap any storage. Wash your hands of that, that's nice. Any storage, yeah. If you get me a few R&D dollars, that's good, good answer. So, yeah, it's good for customers, too. All right, anything else you want to get in before we get to the hunt here? We talked about my two favorites today, but I speak here all day. Soundbite on store once before we break? Store once, still the broadest lineup in the industry, getting broader now with our VSA. And, you know, store once is all about, you know, the most efficient and fastest backup out there still. No one can do it for, no one can do it faster at a better price, so. Awesome. All right, Craig, well, listen, thanks very much for coming on theCUBE. Craig Nunez runs storage marketing, does a, doing a great job, keep it up. Secret weapon marketing. We'll be right back. This is Kadefa this year, and this is theCUBE. We're live at HPE Discover, we're right back after this.