 from the Sands Convention Center, Las Vegas, Nevada, extracting the signal from the noise. It's theCUBE, covering HP Discover 2015, brought to you by HP. And now your hosts, Dave Vellante and Jeff Frick. Hi everybody, we're back. Welcome to HP Discover 2015. This is Dave Vellante with Jeff Frick. Rick Heggberg is here. He is with Sandisk. We're going to talk the All Flash Data Center. Right. That's been a topic of discussion today. We had a customer who wasn't too positive on it, and we had another one who said, it's coming, be here next year, or two years. So, let's start with Sandisk, an interesting company in the heart of this transformation into Flash. You guys making big investments there. You've done some interesting acquisitions. So give us the update on Sandisk. Well, I mean, as you may not know, we're one of the innovators in the nanotechnology between our joint venture with Toshiba. We produce about 50% of the world's nan bits. So a lot of people don't know that about Sandisk, but we are a large producer of nan. As it relates to, as you alluded to, our investment in the enterprise segment, we spend over $2.1 billion in making a series of investments to build up a portfolio of solutions that we can offer to our customers like HP. And those are software acquisitions as well as hardware acquisitions. And most recently, we acquired a company, Fusion IO, last summer, which is the largest of the acquisitions we've made. So it's a big, big investment Sandisk is making to deliver enterprise solutions based on our nan technology. Yeah, so, and of course everybody knows you for the sticks that you've put into your laptops, but you've got a big enterprise play and there's, say, some significant investments there. The Fusion IO acquisition was very interesting. How's that going? Can you give us an update on that? Well, the Fusion IO acquisition has been very good because it brings unique technology and solutions that, for example, our relationship with HP, we've got a very strong relationship with Fusion IO did. We're garnering that as well as Fusion IO has a very strong end user sales organization. So what we do with HP is we actually engage with our end user customers and help them design and promote solutions. So it's not only the technology and the business we bought, but it's also, frankly, a lot of the selling engagements and the relationships that we've had with that acquisition. So it's been going on very well. And Fusion had a lot of software in its portfolio. Where does that fit with Sandisk? We have a, within our enterprise business unit, we have a software team and we sell a variety of different software offerings. These are primarily caching alternatives. So in a variety of servers, you will cache your hot tier data and that's effectively what Fusion had as well as Sandisk had as well. And we're also, as part of the other acquisitions we had, we have other type of scale out software solutions that we have for sand, Cassandra, different types of workloads. So it's a big part of our offer is our software offering as well. We've been talking today about the All Flash Data Center. I said we had a customer from Alcoa on and he said, well, I'm not really sure. He's not really driving data reduction hard. Being careful about that. And another customer on Cloud Service Provider going all in. So let's back up. You're further down the value chain. So you got, you're close to the silicon. Let's talk cost. Let's start there. We're talking raw storage cost today coming out in an array of what? $3, $4 a gigabyte? Is that about right? I mean, you guys are a lot less obviously at yours. With just HP announced yesterday or this week, right? Which is their new All Flash array. They're now with using a four terabyte drive, they're able to now to get to 1.5 for usable space, right? And that's with data reduction, right? Yep. Okay, so, and it seems like flash prices, not seems like as a fact, flash prices are coming down faster than those of spinning media. And what's driving that? Well, it's again, it's solid state technology, right? The fundamental problem with spinning desk, what we call spinning rust is that it's, there's only so much you can do to reduce the cost of that. Versus in a solid state drive, which is basic, it's based on semiconductor technology. We continue to drive down the cost. We put more bits per cell. We drive down the lithography similar to what any semiconductor process is. And so we can continue to drive down the cost. As well as we're actually increasing the capacity into these particular drives. We've just announced a four terabyte drive that we're shipping. Right. And we'll be announcing an eight terabyte drive at some point here in the not too distant future. So we're going to add the capacity, but as well as the cost is going to come down. And the other thing relative to these products at a server storage level, it's also the total cost of acquisition or ownership. Because the power consumption of a storage array and or a server based on flash is significantly less than that of a hard drive. Well, and it goes beyond that when you start getting into data sharing and the ability to spin out multiple copies. And performance is another big aspect. You need less to service the same applications. But I want to go back to what you were saying about the spinning rust. We call it that too. But spinning this is always benefited from Moore's law. I mean it's always tracked sort of Moore's law, but it's begun to attenuate. Right. Because the investments aren't there and head technology anymore and... I'll be honest, I'm not an expert on hard drives, but I can tell you we, if you just look at where we're going with our cost nodes of our technology... You're on that track. We're on that track. We're on that Moore's law track. And disk is slowing down. It's not at the rate that we are. And so at what point do you think the raw cost per bit crosses over? Well, right now we're seeing and you've heard it this week here at HB we're displacing 15K hard drives in server and storage applications. We're now displacing 10K hard drives. And I mean that crossover point's happening now. And over the next couple of years, frankly our objective is we want to replace all that spinning rust at 15 and 10K hard drives and using SSDs. So we think it's going to happen. So spinning disk gets depositioned or repositioned into the bit bucket, I guess, goes compete with tape. Right. Good luck on the cost side. That's going to be an interesting battle, especially when you start putting flash and tape together because tape starts to get interesting performance metrics. But so we're there, we basically were there with 15K, high spin speed disk drives were there. And you're saying 10K within what? Now. 10K now. We're doing it now. I mean you'll see companies like folks here that are going to replace 10K hard drives. And so what's left? 7,200? 7,200. I mean, frankly, there will always be a place for hard drives for archival data to say any other. But when you're talking about tiering data, high performance data, this is where you're going to want. Well, but there's another equation here, which is how long does it take to rebuild a 20 terabyte hard drive when we get there? What are we at today? What's the highest capacity hard drive we can get, do you know? Well, you're four terabyte. Hard drive? Yeah. Four terabyte. Oh, I thought somebody announced something higher than that. They may have. I know. Because you're at four terabyte. We're at four terabyte. Yeah, I think it's- We're starting to, we are now matching the capacity points of our disk today. Okay, so how long does it take to rebuild the four terabyte flash drive? Well, I mean it does. That's the beauty of it. Right. Oh, okay. And how long is it gonna- And the performance, the IOPS performance, everything else. There's tremendous amount of other attributes associated with that particular product, as well as the cost points that are coming down. Yeah, and even forgetting those for just a second, just to rebuild a 20 terabyte, 30 terabyte hard drive, because if you're tracking Moore's law, you're going to get there. It's going to take, I don't know, a month, two months. So what do I do? I got to protect it. Now you got to do things to protect flash as well, but there's no time to rebuild the flash drive. So it's an unfair fight. Now you bring in the environmentals, this notion of data sharing, the fact that I don't need as much capacity to serve us the same, it's- And the reliability. Talk about that. Because that was the knock early on in flash. People used to say, well, if we had to pick a technology to put into the enterprise, we never would have picked flash. Well, thank you, Steve Jobs, for getting the price down low enough that the industry could do unnatural acts. Right. And then leveling and all this other crazy stuff you do in software to make it reliable. People warranty those things now. It's as reliable as spinning disk, right? Is that fair? More reliable than spinning disk. More reliable, so talk about that. Well, because you have less, it's a non-mechanical device. So by, in theory, just that alone, irrespective of all the semiconductor processes associated with it, it's always going to be a more reliable product. As well as we talked about before, it's less, the power consumption alone is significantly less. And that alone drives up the reliability factor as well. So there's just tremendous amount of reasons why flash makes sense in these particular products. So HP, a big customer, you got to be careful, I understand. But Martin Fink, in particular, will talk about the machine. They'll talk about memristor as a superior technology to flash. But it strikes us that cost-wise, how do you compete with this? Right. Yeah. I mean, that seems like it's a long ways away before you can get the volumes out of whatever, some alternative to compete with flash. Now, I know a lot of people out there, like John Luke Shotlain's one, he watched theCUBE, and he's probably tweeting me now, going, no, you misunderstood. But the volumes are enormous. It's classic x86, right? Where the volumes are such that good enough becomes good enough to catch those things. It becomes the best. Exactly, and then the multi-formatted scale of which you can have so many advantages beyond just the unit cost, I think is really kind of the message that's come out today, David. So the other thing you got to remember now is that, there's SATA flash, there's SAS flash. Again, these are higher performance interfaces. So SAS now gives you now 12 gig of an input so that improves the performance. And as you know, we have a line of products based on Fusion IL, which is a piece of express. Now, we move that flash right on, right next to the CPU. We low latency applications, which is effectively moving all your hot tier data there. So we cover various different aspects of your storage requirements using flash. So that's the one key benefit that hard drives could never do, is they can't cover that level, all those different multi-tears of storage capability. So I want to talk about that. That's a very important point you just made. But I want to go back to the notion of alternatives to flash. You set off camera, yeah, we're looking at everything. But what's the boardroom conversation like with regard to those alternatives? Do you generally, I mean, everybody's paranoid. Thank you, Andy Grove. But do you generally pretty confident that flash is here to stay for a while? I mean, you got this thing has some legs, right? Some runway. It's not like you're panicking saying, wow, we got to get out of this business, and we got to shift our... We're just at the inflection point, honestly. I mean, if you just look at PCs themselves, I mean, we're now at about a 40%, 50% attach rate for SSDs and commercial notebooks, consumer notebooks. It's sub 10% attach rate. So again, as these costs come down, you know, and everybody's putting everything in the cloud anyhow, we're going to see the attach rate in SSDs increased significantly in your corporate and consumer notebooks. And then, you know, obviously what you're seeing here today, flash is becoming the predominant storage media for cloud-based infrastructure as well as enterprise infrastructure. What the heck's going to happen to the hard drive, guys? Well, now, I guess, so WCHGST buys Verident. Okay, so they got to play there, I guess, right? Playing Christensen. They use these drives as a setting. What's Seagate doing? Do you know what's Seagate doing? Well, again, they're looking at alternatives themselves, I would suspect, but, you know, frankly, that's the market we're going after. Yeah, I mean, I'm saying, you needed Seagate and other drive manufacturers to do all these unnatural acts that nobody else wanted to do. Fly heads lower and, you know, try to spin it faster and increase track capacities and, you know, just things that, you know, physics would say you can't do, but they did it because they're really smart people. And then they'd sell them for, like, less than your kid's PlayStation, way less, right, with less margin, all right? Well, maybe not less margin, I don't know. But so, yeah, you're going right after those guys hard. Scary. Well, I mean, if you just look at a lot of these traditional old storage array products were based on hard drives, what's happening now? What's the biggest growth market today? He says all flash arrays. And the storage segment all that is the biggest growth opportunity. So I'm getting the- And that's going right after the hard drive. We're getting the time outside, but I wanted to come back to talk about something we call fame, flash as memory extension. So eliminating SCSI protocols, any disk protocol and making flash as an extension of memory, a persistent extension of memory. That's something that Fusion IO, really the first to sort of popularize that notion. A lot of software to do that. Super low latency, orders of magnitude, lower latency than spinning a disk. What are your thoughts on that? Well, again, certain things I can't comment on, but obviously we're focused on high performance flash, which is right up what you just described. And we've got several things that we're developing to continue that momentum in that particular marketplace. So again, we want to play in that hot tier of data as well as we're going to, we'll compete in the low end of the archival data, but that's very important to us. Well, and you're seeing others starting to talk about it now. So EMC talks about DSSD, they made the acquisition there. So that would be competitive to what you do, correct? Or, okay. It's an alternative, mech way to get there. It's a different form factor approach, but it's fundamentally the same type of thing. But similar value proposition. Right, right. EMC talks about CAPI, I don't know if that's an atomic right type of thing, but I think it sounds similar, but you way, way ahead in terms of the software maturity. We are, we believe so. And the contributions that you've made to the Linux foundation in order to improve paging and things that people forgot about, because they didn't need to for decades. So interesting stuff to be continued. Yeah. All right. Well, thanks for your time. Yeah. Thank you very much for coming to theCUBE. Yeah, it was great. Thank you. And I'd love to carry this on at some other time. So it's really hot space and we're excited about it. You can tell we're passionate about the future. And so thank you. Appreciate it. Thank you. Appreciate it. All right, we'll be right back after this short break. This is theCUBE from HP Discover 2015.