 Live from San Francisco, California, it's The Cube at VMworld 2014, brought to you by VMware. Cisco, EMC, HP, and Nutanix. Hi, welcome back to San Francisco, everybody. This is Dave Vellante. I'm with Wikibon.org, and this is Silicon Angles, The Cube. The Cube is our live mobile studio. We go out to the events. We extract the signal from the noise, and we're here at VMworld 2014. This is our fifth year at VMworld. We're in Moscone South at the lobby on the right-hand side just before you get to the escalators. Stop by, say hello. Vish Mulshand is here. He's with HP, Cube alum. You know, I'm going to get to your title. You do so many things at HP. Technical background, you do strategy, you do product stuff, so welcome back to The Cube. It's good to see you. Great, thanks, thanks, Dave. Great to be here again. We were, we spent a lot of time together, last week actually, right? We were in Boston. Yes. Doing the deep dive, you had a bunch of analysts in, kind of doing the Kool-Aid injection. I thought it went really well. You had a good crew there. It was a very interactive session, a lot of good feedback, you know, those are good. I mean, what'd you think? Yeah, I thought it was great, you know, in my mind, getting the perspective from folks outside of HP just to keep grounded in what the reality is, I think is very key, right? And so we really enjoyed the interaction, the feedback you guys provided us, and the depth and time that we could spend on the topics, right? When I was there only for the first day, I told you I was out golfing the second day, it was my birthday, but I got to hear the flash session, and then, you know, there was a little bit of discussion on software-defined, but I think you guys went into that in more detail the second day. So I want to start there. Sure. The software-defined data center, you know, the, what used to be called the software mainframe, you know, they don't use that term anymore, the marketing guys took over from Moritz. And so when I was seeing that sort of instantiation, what do you make of all that? What's HP's sort of position on that? Yeah, so Dave, let me talk from a storage perspective, right, because the software-defined data center is a very broad area, servers, networking. So if we look at storage, there are two elements you want to think about. In fact, three elements you want to think about with software-defined data centers and storage. The first element has to do with cost optimizations. How do you get the lowest cost storage? That's defined by software-defined storage that's hypervisor agnostic, that's hardware-independent, and that is controlled or orchestrated by industry-standard offerings. That's the first piece, right? Then there's sort of what I call performance-optimized storage to deliver on a service level. And instead of, you know, collecting masses of hardware to deliver that service level, a lot of the optimizations are done in software. So as an example, priority optimization software to guarantee how much an application gets in terms of performance, how do you solve the noisy neighbor problem? Or here's another one, pure motion to move data between, say, a flash array and a tiered array, for example, just because it's a different service level you want to accomplish with that offering. So this notion of a service level is really key. And the third piece, Dave, I think is this notion of orchestration, right? And I sort of view OpenStack, you had OpenStack announced at VMware as well. You know, it's the TCP IP of orchestration if I can use that term, right? You know, you want to be able to orchestrate in a standard fashion. Just like, you know, we used to have Decnet, SNA, Apple Talk, and then TCP IP one-out, right? I think you've got that same phase here with orchestration today. Right, okay, so a reliable approach to orchestration that everybody can trust, everybody understands and, okay, so now, so that's kind of the high level. What's your specific product strategy around software defined? Sure, so we can talk about two key products. From a software defined cost optimized, we have the HP Store Virtual VSA and the HP Store Once VSA, right? These are both virtual storage appliances that work on any hardware, any server hardware, right? We, of course, will talk about HP servers, but if you want to run it on a Dell server, by all means. Any X86. Any X86, right? We announced support for KVM on the Store Virtual. We announced support for Hyper-V on Store Once. We also announced the Store Virtual offering being a part of the Helion, HP Helion OpenStack distribution. And if you recall, Helion OpenStack has both a community edition and an enterprise edition, right, and so whichever edition that you get from Helion, you essentially have now the Store Virtual VSA built into it. Okay, so, and we know a little bit about Helion. We've had Sargillay on a number of times on theCUBE and we've seen HP's cloud strategy evolve. So, and we can come back and talk about that a little bit. So, relative to VSAN, I got to get your take on VSAN because there's so much confusion in the marketplace. Chuck Hollister write a blog one day and you'll read it and say, oh, maybe sort of disin the competition. And the next day it's like a lot of love and embracing. But it's clear that one positioning for VSAN is you guys are not just VMware. It's more than just VMware. But what's your take on VSAN? What does it mean for your strategy as an ecosystem partner that sells probably more VMware licenses than anybody? I mean, how do you, what do you make of that whole thing? That's a great question, right? So, VMware continues to be a very close partner with us, right? I think the introduction of VSAN and Evo Rail, I think it just continues to point and validate this notion of software defined storage, right? In my mind, Dave, it's software defined and flash are the two key disruptors. We saw that this year, I think going into next year, we'll see that sort of go even more mainstream, right? So, I think it's great to see multiple offerings here, validating what we've done actually with software defined before it was even called software defined, right? If you look at store virtual and how we offered it. Okay, and so, just a natural progression of the ecosystem, right? And it's like VMware's a software vendor doing what software vendors do, grabbing pieces of the stack and hardware guys got to move fast, you know? Hardware guys that do software got to move fast. Well, I think it's going to be interesting, right? Like any sort of emerging technology is always a flourishing of offerings, right? And I think that's the great thing about this. It's choice, you can say it probably have different approaches and let's see which one wins out in the market in the end, right? All right, let's talk about Flash. So, you guys came out last summer, Flash announcement, all Flash array, based on 3-par and made the statement, okay, well, we're not going to go buy a Flash company, we don't need to. A lot of people, myself included said, well, maybe you don't need to, but Meg Whitman has said we're not doing any acquisitions, certainly any major ones, so you really don't have a choice. So, the question in my mind at the time was, okay, is this a bolt-on? A term that you guys used a lot when everybody was saying, oh, we have thin provisioning too. You said, ah, that's a bolt-on. And you were largely correct. And so, I was skeptical. And then when you came out with Flash last summer, the pricing was, in my view, not competitive. Now, fast forward to this summer. All of a sudden, you're under $2 a gigabyte, your latency is down to best in class. Wow, okay, what happened, how'd we get there? So, where are we with Flash? How all of a sudden did we go from, really, essentially, an okay product with a great stack, that was really your advantage as you had the stack, to one that is now great stack, competitive from performance, and a price standpoint. What happened? Yeah, so Dave, it's been a great year for the last 12, 18 months on Flash, right? If I can roll back the clock a little bit and talk about some of the elements of change, right? I think, to answer your question first, what happened, right? There was a very big emphasis on Flash. We've had R&D developments over the last two to three years focusing on Flash optimizations. There were a lot of skeptics at first that said, hey, wait a minute, you guys are a disk-based architecture. Can you really do Flash? I think the proof's in the pudding right now, right? 900,000 IOPS, 200 microseconds of latency, write latency, 10-day duplication inline, switch data services, data mobility. Now, if you roll back the clock and look at sort of what the ride's been, in December of 2013, we announced something called adaptive sparing, right? So we took now one very key Flash optimization. We took an 800 gig SSD drive and looked at how over-provisioning was done on the SSD drive and said, wait a minute, we can be a bit more intelligent with this, right? Adaptive sparing allowed you to reduce the over-provisioning capacity that the drive takes. So the net effect that the customer was, they got extra capacity at the same price because we treated the Flash differently from, say, a traditional media, right? And so a lot of times I will tell folks, hey, if you're really Flash optimized, Mr. Vendor, where's your adaptive sparing, right? Because here's a perfect example of how I can take the 100 gig drive, deliver a customer 920 gigs. That's 20% more capacity-free, right? So that was back in December where we announced adaptive sparing. This is one of several optimizations we did. Then in June of 2014, we announced sort of Flash for the mainstream, right? $2 per gigabyte. We had a 10-digit application, 10 clones, 1.9 terabytes, CMLC drives, 460 terabyte raw capacity, right? Five-year warranty on the drives, six-nines guarantee. We brought together a real collection of very, very compelling, I think, features that allow customers to take Flash to the mainstream, right? So far, we've seen great uptake on that. We'll talk about customers in a second. We see not only just all Flash deployments, but people at the point, traditional high-end arrays, like a monolithic VMAX, for example, are really looking at that and saying, wow, you mean to tell me I can get the same performance, same residency, half the floor space, maybe less than half the floor space, less power. It's a very compelling proposition. Is that the competition, VMAX, or is the competition other Flash arrays? Well, I think you got both. You got other Flash arrays. You've got also high-end arrays. And then you also have people that are looking for workload acceleration, right? Consolidation, so it is truly becoming mainstream because we're seeing multiple use cases, right? And then fast forward to September, which is just a couple of days ago, we announced all Flash 7200 Starterkit for $35,000, average street price, okay? And if that was Flash for the mainstream, this is now Flash for the masses and my first Flash array, let me use that term. And here's where we're looking at that for days, right? So there are two kinds of buyers, right? One buyer says, hey, I have a limited absolute dollar budget. Say I've only got $50,000. All right, so now you have an offering that gives you that ability to go into Flash. There are also people that are saying, wait a minute, if I'm going to try out Flash in my data center, 35K is a very low-risk investment, right? Maybe it works great. If it doesn't, all right, we'll move on, right? And so I think that's another very interesting approach to the way people are buying Flash. All right, you mentioned customers before, so I was going to ask you, how's the uptake? Have you seen, since you've made the new announcements, have you seen a big boost in demand and getting the proof points that you can share with us? Yeah, Dave, so we've been seeing great uptake, lots of interest, lots of demand. Let me talk about three customers today, okay? So let me start off with Lattices. Lattices is a cloud service provider, they're a managed hosting provider, and they were looking for high performance storage to maintain their SLAs, right? And in addition to sort of guaranteed high performance, they needed the ability to ensure that they could offer customers a consistent and guaranteed performance level as well as a very performance level, right? I may come in with a bronze service level that I need, and I want to pay for a bronze service level versus, say, a gold service level where I actually want to be able to offer that service. So Lattices put the three-part 7450 and then the species software called Priority Optimization to do exactly that. They also use three-part because of its unique multi-tenancy features where you can run mixed workloads, you can consolidate different types of customers on those workloads, that was key for Lattices. And then what they said to me was, provisioning now took hours instead of days. Orchestration was quick, and it was easy, right? That was the big thing for them, it was simple to use. Number two, nuance communications. I don't know if you know the company. Yeah, sure, nuance, they make Dragon. Well, they, Dragon Speech Recognition Software, yeah, sure. I saw a lot of Apple iPhone series. Local company. It's on the back end. Yeah, local company. Local for me. For you, for you. So nuance does speech recognition software and they actually help, in the case of Apple iPhone series, non-native speakers, right? By recording their voice patterns and then helping recognize those voice patterns, right? Especially if you're a non-native speaker. Now, they use that 7450 to index those voice files very quickly to deliver iPhone series service. To improve recognition. I mean, you remember when iPhone first came out, the Siri was awful, you couldn't even use it and now it's so much better. It gets better over time, right? Yeah, yeah. So that's a great use case there. Third one is exact target. And, you know, exact target is a marketing demand generation company. They have a huge number of databases. In fact, some of the stats that they shared with me was four trillion rows under management. They do 21 billion rows a day. 100 terabyte databases are not uncommon in exact target, right? And so they have multiple three-power arrays to store this data. And they deployed both tiering with Flash as well as all Flash arrays, right? And the biggest thing for them was how do they adopt Flash without ripping and replacing their infrastructure, right? They have an existing infrastructure. They want to be able to add Flash to it to accelerate performance, lower cost. And then they also now are doing all Flash for VDI, right? So exact target is a perfect example of a three-power customer being able to extend and embrace Flash without doing a lot of change. Three-power, the gift that keeps on giving, I always say. All right, Vish, we have to leave it there. Thanks very much for coming to theCUBE. Always a pleasure. Thank you, it's a great talk to you. All right, keep it right there, everybody. We'll be back with our next guest. This is theCUBE, we're live from VMworld 2014 and we'll be right back.