 Live from Las Vegas, Nevada, it's theCUBE at HP Discover 2014, brought to you by HP. Okay, welcome back everyone. We're here live in Las Vegas for HP Discover. This is theCUBE, our flagship program. We go out to the events and expect a silver noise. I'm John Furrier, the founder of SiliconANGLE. I'm with Dave Vellante, co-founder of wikibon.org, chief analyst of the Focars. We're here with David Scott, man of the hour, senior vice president of storage, general manager, HP Storage up on Keynotes. So much demand of bringing up another keynote. Welcome back to theCUBE again. Thanks, John. Great to be here. Every time I interview you, you're always like the star of the show. Every earnings announcement, oh, three par. So, as Dave always says, it's the gift that keeps on giving at HP. So, tell us, what's hot today? I'll mention a lot of the Keynotes. Why is all the action on HP Storage right now? What's the hot button? Well, I think we've just introduced another revolution in the storage industry. A lot of people have been using all-flash arrays, solid state drives, but really for niche application usage today, and the question on everybody's mind is, when will the tipping point occur? When will we turn around and see all-flash arrays able to replace traditional high-end tier one arrays, hybrid arrays, as mainstream enterprise application usage? And yesterday, we announced the capabilities that allow that exact thing to happen with our all-flash 7450 array. So, it's exciting for me because when I started working with Dave almost five years ago, storage was certainly changing, and flash, we were there from day one of the whole revolution of flash. It continues to be hot. We've been at all the shows, we've heard all the fun, and it's good to get the word out, but I want to just share with you, a lot of folks throwing in daggers here and there, shooting bullets at HP. Oh, HP Storage, what's going on? They don't really have a flash story. It's been kind of the fun. So, I want you to talk about that comment, and then why you guys are so hot right now in flash, and how do you talk about it? Because this is your time to shine. Sure. Well, if you look at the all-flash market, there are three things that really need to take place to allow you to replace mainstream enterprise application deployments on high-end hybrid arrays. It's, first of all, it's got to be fast. You've got to be able to deliver kind of low response times, predictable response times, and that's exactly what we do with the FreePass 7450. 900,000 IOs, less than 0.7 millisecond response time. The second thing you've got to do is make it affordable. The reason why all-flash hasn't taken off is it's too expensive, much more expensive than standard high-performance hard disk drives. And with the announcements that we made yesterday, we introduced a new 1.9 terabyte consumer MLC drive plus inline deduplication technology we call Fendi duplication. And in combination, we can now support a price point that is less than $2 per usable gigabyte. It means we've reached really the tipping point where it's more inexpensive to choose an all-flash array than it is to choose an array with spinning disk. And then finally, we've made it enterprise-class-scalable and resilient. We have some new, unique, patented Express Indexing software in three-part that allows us to put 240 drives or 460 raw terabytes of capacity in a single array. And with Fendi duplication, we get between 4 to 1, 10 to 1 compaction ratios. So that 460 terabytes turns into a 1.4 terabyte scalable platform, really true enterprise scalability. And of course, the benefit of the three-part architecture is it's proven it's always been mission-critical tier one architecture. And to make people feel safe about going into the all-flash world, we've just backed it with five-year warranties for our new MLC drives and a Get 6.9s guarantee program, i.e. we're guaranteeing that in a quad-controller configuration, you'll have 99.9999% availability. That's what makes us the first all-flash array ever to be suitable for mainstream enterprise applications. So one of those things we talked about right at the events we've been to, the OpenStack, a lot of the big data events, is it's got a lot of unstructured data. There's a huge tsunami of unstructured data coming in the market. Also, you have structured and for transactional like processes. And a lot of people are the high priests and the smart guys, and interesting, there's no one-trick pony for flash. Meaning you've got to address the IOPS throughput and cost per gigabyte because not everyone's storing unstructured data for low latency. So you have to have the breadth of use cases. Talk about that dynamic. Is there a one-trick pony for flash, or do you have to be multi-purpose in the sense of having the performance at those three levels? Well, I think in a sense, if you've looked at the niche applications that flash went after initially, you had people going for high-performance database optimization where low latency was really important. And then you had some of the flash startups focusing on trying to be more general purpose, but really their systems only had the resiliency that could sort of small to mid-sized businesses and scalability. But if you want to go into the mass of the multi-billion dollar market space, you really have to have a platform that can do everything. And that is what we've delivered with the three-part store serve 7450. A platform that provides optimized solutions for databases, for the ability to store files, for the ability to support server virtualization and VDI, bringing it all together in something that enterprises can really trust and rely. Have you heard about some of the warranty issues around flash? A lot of people were initially considering flash, okay, they'll overpay, but there's some media issues around reliability. Is that going away? And what do you guys have done on the reliability side with the warranties around there? You're still maintaining your levels of service? Well, actually, you know, we've now had more than kind of four years experience with flash storage in our three-part arrays. And because we have our very sophisticated remote support capabilities, we can actually understand what is happening to those flash drives from a web perspective. And our architecture has always been highly optimized for flash because we use system-wide striping. We have various techniques that optimize or rather minimize the amount of wear that occurs on those flash drives. And that's given us the confidence even on these new consumer MLC drives to be able to deliver five-year warranties. And being able to put our money where our mouth is, make sure people feel confident that they can deploy all flash and have their data safe. Yeah, five-year warranties in six nights. I mean, when you talk to customers, they're worried about cost and they're worried about the reliability of flash. So you're putting your money where your mouth is there. A lot of people will say, well, you said, hey, we don't have to go out and buy another all-flash array company, like EMC did, like IBM did. We have a flexible architecture that allows us to optimize for flash. We called it polymorphic. It was polymorphic. Polymorphic, I love that. Okay, so there was an interesting debate still going on in the industry. Oh, he says only time is going to tell. What gives you confidence that this is not a bolt-on, as you used to say, within provisioning to some of your competitors? You said it as a gentleman. You're very good when you talk about your competitors and you smile as you rip them apart. But what gives you confidence that this is not going to play out like some of your competitors bolting on features that you guys popularized with flash? Well, so first of all, if you look at some of the flash storage startups right now, they really don't have the enterprise scale or resiliency. Someone who's particularly vocal right now, pure storage. But let's look at an environment where you're in mainstream enterprise application usage. You want a petabyte scale array. On a 7450, as I just described, you can support 1.4 petabytes. But what's the pure storage penalty to meet the same need? First of all, you're going to need six arrays that you have to manage to get to that 1.4 petabyte scale. You're going to have a price point that is twice as expensive as a 3.7 4.50. They haven't been able to break the price point of $2 a gigabyte, which is really the key of meeting the price point of high performance hard disk drives. And you're going to have a worse platform that doesn't have quad controller, high availability. It doesn't have synchronous remote replication so you can guarantee no transaction loss if you have a system failure. There are whole hosts of sophisticated HA capability that simply doesn't exist. So you get a worse solution for higher costs and more management overhead. Now, me for one, having been CEO of 3PAR and taken a private company public, that is not a value proposition that I'd like to take to investors on a pre-IPO roadshow. So the big thing about 3PAR that I see, to me the number one advantage, you got to, you know, the cost is a check by, you got to be there or you're not going to sell any. You got to have the reliability. To me, it's the stack is what you bring to the table. And if you can meet those other two, then I think you're going to do some serious damage. So talk about the importance of the stack. Why is it that you have the stack and others don't? Well, the advantage we have of the stack is that even though we're a modern architecture designed for this century, we have had the experience of kind of 10 to 12 years runtime. So we have a very high quality proven resilient foundation. And we built functionality that allows us to integrate into all of the leading environments, whether it VMware, Microsoft, into SAP, HANA, all of the key environments that anybody who is a large enterprise customer will want to deploy. And so our snapshot integration goes into platforms like Oracle, amongst other things. And that comprehensive solution stack on top of a really robust infrastructure that we can provide with three-part is what allows us to deliver mainstream enterprise application support. Yeah, it's hard to build a hardened storage stack. Okay, I want to talk specifically about ExtremeIO. EMC says architectures matter. We went out, we bought the best architecture out there. How do you respond to that? Yeah, well, if you look at an ExtremeIO architecture, you know, it supports a remarkably small amount of storage capacity. I can't remember, is it 20, 40 terabytes? I mean, on a single system, we support 460 raw terabytes. But they would argue a scale out, right? They would argue a scale out. So you're already at a scale out of 20 before to basically match. So now you've got 20 devices. So here's your complexity, there you are. Starting to manage. Yeah, so it's a real challenge. And it has none of the enterprise scale resiliency built in to the architectures at all. In fact, I felt it was quite a surprising choice, actually, for an acquisition to make, because ExtremeIO wouldn't have been up there on my list. But you would buy VMAX to get that stack, which leads me to my next point, which is, to me, this is a redefinition of what tier one is all about. Do you agree with that? Yeah, that's right, because if you look at where people deploy mainstream enterprise applications, if it's not on HP 3.0, it's going to be on one of our kind of major high-end competitors. It's going to be something like EMC VMAX running in a hybrid auto tiered configuration with flash disk drives, maybe 5% hard disk drives. But if you look at the VMAX tax that that costs, in order to, let's say, deliver 250 usable terabyte comparison with a 7450, you're going to pay four times as much for a VMAX 10,000. It's going to take 25 times more space in your data center, and it's going to take eight times more power to deliver less predictable performance, because it's using auto tiering between disks and hard disk drives, rather than disk and solid state, rather than all flash array. So again, the difference between that is so huge. In fact, I think a comparative difference that I could point to would be back at the start of the 20th century when you compared buggy whips with automobiles. I mean, it's about as chalk and cheese as that. So I mean, the fundamental issue is for everybody who is considering refreshing their VMAX technology now, there is a better alternative. There is a mainstream, all flash array with HP 3.0 that can meet their needs better than anything that EMC can provide. All for that matter, are the vendors like IBM or Hitachi? I want to pick your brain as a Silicon Valley CEO. A lot of times I think 3.0 doesn't get enough credit for what was built. And I think the reason was that the buyout happened so fast, you know? And so, but 3.0 and data domain, and maybe some others, but those really stood out. And now you've told me a number of times that the amount of money that you had to raise to be successful was enormous. I forget what the exact number was, but it was. We raised about 183 million when we were private. So 183 million when you were private, and then you did the public offering. Another 90 million when we were in public. So you see pure with a $3 billion valuation, they've raised 250 million. Just in the last round. In the last round, right. And so in total, it's north of that. 400, 500 million, something like that. So is that the right approach? Because they need that more just to take others on. Or are these valuations a little bit ahead of their skis? Well, I got to tell you, I think it's an incredibly rich valuation for a company that is still apparently making losses. Otherwise you'd expect it to have kind of gone to be public. And you know, one of the interesting things about the startup industry is that you really need to have more than the average two year head start that a startup has. When we were three part, we actually were very lucky because coming out of the gate we started launching into a recession. And a recession has a really interesting effect because it takes away other startups who aren't well funded. So you have less competition. And it also cuts the R&D investments for some of the major storage companies so they can't catch up with the innovation that you bring to market. And so your time to market advantage balloons from maybe two years to four years. And that four years tends to be enough to get escape velocity. But if you don't have a recession and you just have the two year advantage, you're always susceptible to a major company if they have the right architecture coming back at you. And that's exactly what has happened in this case. The all flash startup array vendors emerged around two years ago. And literally with today's announcement with a three part store serve 7450, we have really put an end to the need for niche flash siloed startups. So this is one of the more interesting answers we've heard to is there overvaluation in the flash. Your premise is they won't have the escape velocity to get out as far in front of you guys and the EMCs and the IBMs and everybody else as you did because of that dynamic of the dot com bubble burst. And I actually don't think that there's any way that they could get into a public market at a $3 billion valuation at this stage of the company. It's going to be a very, very long wait now for them to be able to catch up to that valuation. All right, we got to end it there, David. All right. David, thanks for coming on. Final question, share the folks in your own words. As we've interviewed you over the past multiple years, lost change, what's been the significant change in short of the past four years in storage? What has been the big, big shift and that you could digest down into a soundbite? All flash arrays arriving. Big time faster than you thought? Big time, okay. David Scott, Senior Vice President General Manager of HP Storage, changing the game here. Always continue to move the needle, big growth engine for HP, great to have you on theCUBE again. We'll be right back after this short break. Thanks, John. Thank you, David.