 Hey, welcome back to VMworld 2013. This is theCUBE, our flagship program out of the advanced extracted city from the noise. I'm John Furrier, the founder of SiliconANGLE. I'm joined by my co-host Dave Vellante, co-founder of Wikibon.org. Go to Wikibon.org for free content. Go to SiliconANGLE for the reference point for tech innovation. And go to SiliconANGLE.com for all the footage. Also go to YouTube.com for the SiliconANGLE for all the replays. I'm joined by my co-host. Hi everybody, I'm Dave Vellante. Lee Caswell is here as the Vice President of Virtualization Product Group at FusionIO. Lee, welcome back to theCUBE. Good to see you. Thank you very much. It's great to be here. See you guys again. This venue is terrific. I love it. Yeah, you're in a new role actually, right? New company, new role, it's a very exciting time. So how's it going for you? How many VMworlds have you been to? Oh, yeah, since the start, right? Are you a 10-year veteran? Exactly. So, yeah, I've seen a lot of change. And, you know, since virtualization, I mean flash is the next big exciting thing. So 10 years, I mean a lot's changed. First five years, just give us your perspectives. You worked at VMware, right? Yeah. First five years, and second five years. Just what's the summary? What's the bumper sticker? You know, when we started off back in like 2000, we basically looked and said, well, what are we going to virtualize first? And it was the easy stuff, right? Take all the applications that were running that weren't very IO intensive. It wasn't the Oracle databases we wanted to go put on virtualization. Now, you've got what, 70, 80% of workloads being virtualized, what's left? Well, all the hard stuff, right? And that's where flash is coming in is how do we go and take the hard applications and make those sing in a virtual environment? So Austin, you're heading up the virtualization team at Fusion, is that correct? That's the new roles, that's the official title? Yes. So what's the big news for you guys this week? You know, we've got some very exciting deliverables that we've shown. We have a technology demonstration we're doing on a new product called IOVDI. IOVDI basically solves the problem of how you get performance into virtual desktops without breaking persistent storage and giving you a cost that's less than a physical desktop, which is what everybody wanted from the start. So you have to solve the cost problem, solve the performance problem, IOVDI is that. That's the latest import, the latest implementation of our IO turbine software. So it's a very interesting way to go and say, we'll take all the benefits of the IO memory flash platform, which have been the basics of Fusion IO success. So you and I had a chance to chat on last week prior to the embargo of the news, but one of the things that we were talking about, and then I was talking with Dave about earlier this day was that, everything at the top of the stack has always been this elusive dream, right? When Paul Moritz laid out the original vision in 2010, it was really laid out, we called it the software mainframe, whatever you want to call it. It was a stack, and at the top of the stack, you had apps, VMware tried their hand at that now, Pivotal's outside, and there's still a lot of work to do in the middle ground, right? So I would say it got stalled a little bit, mainly because the hypervisor stuff, a lot of the middleware, big data hit the scene, storage virtualization, network virtualization all kind of started to happen. Yes. So with that, what's happening above the stack? So stuff's starting to be commoditized, infrastructure service, platform service, but then the apps, data fabrics are there. So what's, you're at the top of the stack, you got to look up. What's the view and what's the trends there? Well, one of the aspects of virtualizing flash is that we're looking at basic hypervisor level virtualization first, and this was the phase one of what IoTurbine had to develop, which was how do we go and solve the IO Blender problem? So any virtual appliances or virtual machines have to go and look carefully at how we're going to go take what looks like now a random workload and how do we accelerate that? That was phase one. Now, we have with IO VDI a very interesting way to run in the guest and add more intelligence. And so the intelligence now could be, hey, in a desktop environment, how do I take advantage of common files to speed up boot times? How do I take advantage of the fact that there's a substantial amount of desktop writes that actually never matter? Remember your desktop, even that drive goes on, you're like, what's it doing? That's all data that doesn't ever have to go to the same. We could take advantage of this now intelligently at the guest and do some very interesting work to speed up acceleration, make sure desktops are working fast. And that's the sort of intelligence you look at, and it's all based on applications and solution knowledge. One of the things that I've been working on at Fusion IO. So I got to ask you, Lee, I've been coming to VMworld now, probably, I don't know, six or seven years. And I remember my first VMworld, I said, oh my gosh, storage is going to get killed, right? And it was everybody's complaining about storage. And so then we started on this path of integration. You know, VAI and VASA and the like. And every year Wikibon does this evaluation of the integration points and we rank, who's got what and I'm looking at the other day and I'm saying all this stuff is designed to sort of minimize the spinning disk penalty. And I'm looking at the integration points that relate to flash and there's like a handful of them. So to the extent we get to that vision, it seems to be coming soon, where all my active data, we talked about this with Gary earlier, while all my active data is served out of flash, all those other integrations that I just spent all this time and money on kind of become irrelevant. That was my take. So the first time I've articulated that, I wonder if you're an expert in this area and products, is that a fair characterization? Yet for years, the disk drive has been doing a dual service. It's been providing both performance, which it's not very good at, and capacity, which is very good at. And so what's happening is that as you look at flash right now, and this is one of the reasons Fusion IO is so successful early on, is a single PCI card serves the performance delivery of over 200 drives. And so what's happening now is there's this radical split happening where wherever you can take the performance and disaggregate it from the capacity needs, now that's changing extremely fast. And so we're seeing that overall where I'm going to use a disk for a relatively cold store, anywhere I can provide acceleration, the software stack is how we do that. Yeah, well if I could do that through an API call, based on some kind of policy, so where are we in terms of being able to do that, and what role does Fusion IO play in that regard? Yeah, very good question. We've done some very interesting things with IO control, for example. This was an acquisition we had recently where we're now applying quality of service across as a policy across application environments. So if you want to have a SAN and basically run multiple applications, how do I go and make sure that I've got performance now that I can allocate so that I can make sure that I'm getting the performance I need for the applications I care about? Allocating not just baseline performance, but quality of service becomes a very important differentiator that Fusion IO is driving. Okay, and I can do that through an API call. That's right. It's an open API. Yes, and you can go and actually allocate this on a policy based by your application. And I can change that pretty much on the fly? On the fly, yes. It's one way of thinking that it's not just raw performance that users care about. It turns out what users care about, and you know this from your own experience. Waiting for that little hourglass to change. What you care about is you care about consistent, or let's see, consistent performance as much as you care about absolute performance. Predictable, consistent performance, right. Yeah, the one thing that drives users nuts is if they don't know when something's going to complete. Right, and if it's too slow, then they'll throw it out and get a new one. But if it's consistent and predictable and I know what's coming, then I can build processes around it. Here's one of the areas we've been spending a lot of time on. We were so early with Flash. We spent a lot of time on solutions. So if you look at what are the key solutions at Flash Accelerates today? Well, it's databases, server virtualization, VDI, big data. If you take those as a group, we have a set of customers that have deployed and seen successful acceleration in the field. And we're just going to show other customers, here's how you can do this. We've stripped out all the risk of making this work in the field. So talk a little bit more about the customers and how the use cases are expanding, kind of where they started and where you see them going. And I know there's a wide variety, but I wonder if you can generalize it, especially as your product line has become more robust. Well, we've taken a mapping right now of whether you're on a server side, are you on the storage side? With caching, are you going to basically try and bridge the gap between these? And the applications look like this. So within databases, databases love block storage and they love fast response times. You can service more customers. You can save costs. You can consolidate infrastructure. These are terrific benefits now for how Flash can make a difference. In server virtualization, we've got the ability to go and run more VMs more consistently. That's a huge driver of getting more virtual workloads going. Virtual desktops, got the same concept of how do I make sure that users get that level of consistent response times. And then lastly in big data. Big data is all about processing. No data is deleted anymore. The data that you have is just processed over and over and over again. And that processing is all consistent with high performance Flash. So big data, you're talking about extending in memory analytics potentially, persisting in memory analytics, right? We have some of these HANA crazy, but HANA needs to persist that data. We've been doing a lot of work on HANA lately. It's great. We love the concept, but you talk to HANA users and they keep telling you, well, it goes down a lot. You need to persist that. I know you guys are working hard on helping SAP out with that problem. Well, there's some very interesting applications. We announced Spotify as a customer, for example. Streaming music is an ideal case of how do you have very fast performance over latency sensitive applications. These types of things and how you go and manage things like playlists, right? Become very important for businesses that want to take all of the effort they were doing on managing IO. Take those developers off that work. Put them on developing new applications or new features that you're going to use to compete against your competition. That's how you've changed the game right now, is I don't have to actually worry about managing IO because we have thousands of IOPS to work with. Hundreds of thousands of IOPS. All of a sudden, what was a scarce resource in the past? Now you've got a lot of it. So think about re-architecting. That's the sort of cathartic change we're going through right now. Lee, how do you talk to guys? First of all, there's two questions to this one. The first one is, Silicon Valley's always a new startup's coming along. So like, are there any seats left at the table in the IO game? We'll get to that one in a second, but I want you to get to the second one first. Which is, if you're an IT guy and you've got all this storage laying around. Yes. You know, NAS, SAN, DAS, all of this laying around. Usually tied to some app. By going server side, talk about the dynamics that you guys get in there. Is it a rip and replace? Is it an extension? Do you guys commoditize it? Is it just, you treat storage as a resource that can be commoditized? I mean, how do you view that? What's the solution? It's very interesting. Well, one thing we're finding is that there's so much extra capacity now because customers are buying disks to deliver performance. That element, right? Having to buy it. So, you know, a 15 case S drive gives you 150 IOPS. It costs $7 to get that level of performance. Flash is relatively inexpensive at a nickel. So you can all of a sudden now, you can free up all of this capacity. So, one of the things we're seeing first off is what drives buying decisions is how do I consolidate the infrastructure I have? We're consolidating physical infrastructure. We're consolidating licenses as well by having this level of performance. So that's one dynamic. Customers come in different shapes and sizes. Some customers want to buy server side flash. Some customers want to buy storage side flash. We're delivering both. We have with our ion products and IO control products. If you want to buy storage, we have some very interesting ways to deploy it that way. If you want to buy servers, we've got the fastest in the industry on the server side. So, you know, our mentor right now is, however you want to consume it, we're going to supply it. So the economics is you can come in and maximize pre-existing investments. Same time, get that flash data center built out. Is that kind of like? Yeah, let me describe one way we're doing that with IOVDI, which is new for virtual desktops. We're coming in saying, we're taking all the performance dependencies from the SAN and basically moving them into the server side. So by having it on the server side, now you can say, well, I'll just tap into the SAN for capacity, which is really what you wanted in the first place. I just wanted that SAN for data protection. And so the SAN administrator says, great. This is what I was hoping to do in the first place. Give you a few terabytes. You're off and running. I deploy this on server side deployments. Basically gets you back into that seamless increments of deployment. Well, we saw a lot of action today in the news. Violin filed to go public. We love that. So competition there. There's always new startups coming out. So what are you, back to the startup question. There's always a new startup, IO's hot, so you see some innovation. What are you seeing on the startup scene and are there any seats left at the table? Well, who knew storage was going to be so sexy? Well, we did, but I don't know. I guess you guys did, right? John Furrier is the co-decorator of storage. You see it, really? And Joe Chuchu just agreed with you, John. I asked Joe Chuchu if storage was sexy. I'll tell you what, you know what I mean? I don't know if it's sexy, but he turns out he does. He's in a taut. It's funny. I mean, there's a lot of room for innovation left. This is what we're seeing. Flash by itself is one way to go and deploy this. There will be others right over time. What we're looking at is, once you take any imperfect media, and Flash like disk is an imperfect media, you have to start thinking about, hey, how do I basically overcome some of the limitations? There's reliability considerations. I got to make it reliable, right? There's density. How do I go and aggregate it together? There's protection. I mean, all of these things, and so all of that tends to lead towards software innovation, right? Software innovation is where we're putting the bulk of our effort right now on making Flash more differentiated. So everybody wants a PCU, right? I mean, you guys, you had like a four-year lead on the industry, and you did the side field. You was like, oh, wow. Maybe not just the Flash and the Pan. And so now all these big guys investing, buying companies, et cetera. So you said software is where the innovation is. Is that how you keep your edge? I wonder if you could talk about that a little bit and help us understand what we can expect generally. Yeah, that's a really good question. There's no doubt, and I've had experience in the past. At one time in my career, I was selling silicon to Intel for 60 point margins. And the question was, so how could you get away with that, right? Those are the days. Thanks, 82. And the answer was. 84, 85. The value prop was not about the silicon itself. It was how did you prove out things like compatibility, software value add, and in our case at Fusion IO, solutions. What we've done and what we offer to customers is, it's not so much about like raw acceleration because anybody can pull a number off a data sheet and say, hey, we're faster in this one case. What we can show is, we've made these customers this much more successful in the field. And so our value right now is to show that we're going to accelerate your success with Flash, not just accelerate some portion of your data. So what are those solutions? I mean, you talked about them briefly before, but so what, it's talking about generic terms, database, you know. Yeah, you bet. It was interesting actually looking, we have a luxury from a marketing standpoint of saying they're actually fairly definable. So, within the database case, Microsoft SQL Server. We've got Oracle, both for Rack, for Oracle 1112.x, MySQL. If you look there, when you look into virtualization, well clearly we've got VMware today and then moving to Hyper-V, right? Within VDI, so it's both VMware for View and Citrix. And then within big data, we see some very interesting work there. And I'd like to comment on that for a minute because of our success on Flash, just showing the raw performance, then we had application developers saying, hey, I'd like to rewrite the applications now. And so we've had some very good success with companies like SkySQL, MariaDB, Percona, of rewriting the applications now to take advantage of the native benefits of Flash. And so that's a tool. It's a magnitude performance potential. It's a very interesting dynamic, right? So, okay, so that's, and that's always been fundamental to your strategy and a big part of differentiation. And you guys are kind of unique in that area. Well, at some point, we're moving from the early adopters. So early adopters, right, they like words like visionary, disruptive, groundbreaking. This is going to be like, well, to the later adopters, right, the CIO of a grain company in the Midwest, like that sounds pretty scary. So what we've done now is we've reduced the risk saying, hey, you get these benefits. And one of the things we have, we have a theme saying, same planet, different world. And that is designed around the aha moment that occurs when people realize, are you kidding me? 40% of our customers see more than 10X performance in their applications. 10X in the field from our surveys. 10X performance, can you imagine the moment where you go, really? Seriously, I could do that. Well, the norm is to get that low latency, you know, mobile feel like, hey, you know, no disc at all. But, you know, I think that's the key. So I want to ask you a final question as we wrap up this segment. Sure, please. So you guys obviously are doing great. And we were talking earlier with Gary Ornstein and some other folks. The stuff under the hood is where all the actions in the data center. So I'm going to find data centers, not just one thing, it's a bunch of parts. Flash is a big part of it. What is the big takeaway for folks out there? Share it, I'll give you the last word. Share with them in your own words. What's going on with Flash this year at VM where all this 10th anniversary? So Flash, the benefits of Flash are so compelling, it's going to be deployed everywhere where disc has been deployed. When you think about it that way, all of a sudden you look at the server side, you look at the storage side, and you look at how you bridge the gap in between, we're going to see Flash come on to everyone. And what Fusion IO has done is said, we're going to be able to give you solutions. However you want to consume it, we'll give an offering there that you can go and take the advantages that we've developed in hardware and software, take that and deploy it at low risk. Final question, just to add one more. You've been a VMware veteran, your industry vet, been around the block, you've seen the movie a few times, kids going to college, our kids going to college. But you've been to all the VMworlds. What can you share the folks from the beginning of the first VMworld to now? 10 years, what has happened? How big has it become? Can you give any order of magnitude, share some perspective or experiences? Sure, in the early days the question was, hey, there was a customer question of, virtualization, is it safe? Right, just to start off with, will my data, will my apps run? And so you go through that first phase of jumping in the pool. Am I going to jump in, is it okay? And then you jump in and you're like, wow, that was pretty good, right? One of my experiences early on was that the first benefit was about consolidation because that drove cost improvement and then the subsequent value was around high availability and management. We're seeing the same thing in Flash right now and you're seeing everyone get in the act. The first element is, hey, is it safe? Is it going to work? How can I consolidate infrastructure? We're going through that, we've gone through that phase. Now it's about how do I manage this? How do I make sure it works in the applications? How do I get HA? How do I support VMotion? These are the questions customers are asking. It's an integration question. We think we're in a great position to capitalize on that. Lee Cattle of Fusion IO, thanks for coming on theCUBE. We'll be right back with wrap up after this short break. Day one, I'm John Furrier with Dave Vellante. This is Silicon Angles theCUBE here live at VMworld in San Francisco. We'll be right back after this short break.