 This is SiliconANGLE.TV's coverage with Wikibon.org. I'm John Furrier, founder of SiliconANGLE with David Flynn, the CEO of Fusion.io, and I'm joined with my co-host. I'm Dave Vellante of Wikibon.org, and David, welcome back to theCUBE. A long time guest, it's great to see you again. Well, thank you, glad to be here. So this is just past the one year anniversary of you guys as a public company. You just had a phenomenal earnings report. You must be really, really pleased. I know you're not done, but looking back on the year, you got to feel pretty good about it. It was a very fun year for a first year as a public company, and 82% growth and acquisition of ioturbine, phenomenal growth in the company, now closing on 700 employees, 700 people. I still remember back just six years ago when there were six of us. David, you were on in 2010 when we did the first VMworld three years ago, and you guys were a privately held company. You went public, and boy, with the naysayers wrong about Fusion I.O. I remember the post that we wrote in context, so they're going to have one customer, and we pointed to you guys as really the disruptor in that market, and that a lot of the companies are going to look like Facebook, look like Apple, and Flash is going to be a big part of that disruption. So yeah, all the naysayers kind of been proved wrong. So my question for you is, well, congratulations. My question is, what's next on the disruption side for you guys? Obviously, you guys had great performance financially, success on the marketplace continues. Flash, obviously, is now mainstream in terms of everyone's mind around, it's obviously a disruptor. But what's next? What's the next disruptor that you see? You know, the most interesting part of this is what happens now that you can use standard server platforms to perform your higher order storage functions. We talk about this as software defined storage. With the miniaturization that Flash allows, you can embed within off the shelf servers the reliability, the performance levels to where it's just a question of software. So the release very recently of our Ion data accelerator, that to us is very interesting because now the primary complaint with Fusion IO just a year ago was, oh, it's local to the server. That limits the applications that it can apply to. Here we have now the ability to do the storage networking capabilities from within standard servers. So obviously disruption with Flash is actually key. I just put up a post on siliconangle.com, there was a guest post, it was called the Cloud Laundry Mat, talking about the conferences and how people are doing a lot of cloud washing. But we've been following cloud, we love the cloud. So we were pro bias towards the cloud. Obviously relevant. But the migration to the cloud is not happening in the mass tsunami. People thought it would be everyone migrating over. On-premise is still a big part of the architecture. And you guys have shown that with Flash that you can do new things. Did that surprise you or, well do you agree with it and did it surprise you? And how do you see new experiences coming out of that? The interplay of Flash with cloud is kind of interesting because on the one hand, it allows for people to do more with their own infrastructure. When you can get a server that's now capable of doing 10 times the workload it did before, you can solve the scaling problems that used to require that you take it off-premise to the cloud to do. So it lets you to scale in your own organization a little bit better. On the other hand, it does allow those who aggregate lots of demand when they use Flash, like salesforce.com who's a customer to provide new types of services, analytics and stuff that weren't really possible before. So Flash applies in either case. It's really about how quickly the organization can learn to take advantage of the performance levels, the ability to analyze so much data, more data than you could have with the disk. We were talking to a practitioner and I was at the Cassandra summit. It was a big data event, which is the alternative to HBase, you know, normal cool geek, alpha geek kind of event. And we were doing the cube there. One guy said to me, he goes, you know, when the camera was turned off we're replacing them, you know, a full sand, EMC sand with, it's completely gone. We use caching on the front end and use cheap commodity disks and to replace a zillion dollar EMC order. They cut it in half from two sands to one and completely re-architected. Are you seeing that too? Are you seeing where these new architectures are being developed? Can you point to specific use cases that you've seen in production where the geeks are in there re-architecting this new era of data center and architecture? You know, we do see that, especially around these performance optimized storage arrays are now no longer necessary when flash in the server can do a much better job of supplying the performance. So on our website, we have literally dozens of case studies. One of the most recent ones is with Polaris, the same guys that make the, you know, the sport utility kind of four-wheelers and snowmobiles. They're using Fusion I.O. to accelerate their virtual server farm and show that their customer's web page load times go down from five seconds to under a second. So a 5x improvement in the responsiveness of the web page. So these kind of case studies that show repeatedly where it no longer requires a performance storage array, you can have the storage array more focused on the capacity problem, which makes it more efficient. Have you seen any use cases where, okay, five seconds to mil a second, so whatever you said, it's got to be a quick pop. How about like the use cases where you got this massive data warehouse of an SAP implementation or something to the effect where you got a customer who's traditionally using data warehouse, where it's days, hours, to minutes? Yeah. You seeing those kinds of examples? Those are probably one of the most pervasive set of examples within our case studies is where people have gone from query times to do a complex query against a large data set and they had potentially been using custom hardware for the whole thing with this dedicated system, being able to use an off-the-shelf server with terabyte scale flash and have those analytics queries go from minutes or even hours down to a fraction of that time. Because you have to think of it as with our architecture, it's as if the data were in memory because we've integrated flash as a new type of memory as opposed to integrating it behind the storage controllers. So one question I know Dave wants to ask a question but one final question is more of a kind of a step back philosophical question as the CEO of this company from a privately held company to growing public company. Are you surprised by how much big data's been winned at your back in the market? I mean, because you guys were doing well just on the classic disruptions vector, flash as an architecture, performance, economies, economic advantage, but now with big data, has that been, are you surprised by how big that's been as the wind in your back? You know, it has been bigger than we expected. The other tailwind behind this is the faster processors. Faster more cores means it takes more data to feed them. So one is the appetite for large quantities of data with rapid access and the other is processors that are capable of actually crunching on that kind of data if you can get it to them fast enough. So those two things together are what actually, the demand for larger data and the ability to process more data accentuates the problem that we solve of being able to supply large quantities of data fast. The bet on Intel is a good one and you've seen other companies. I've been to Data Domain for example, another company that bet on Intel helping be a wind in itself. You're talking earlier about the performance increases and David, I think back to the conversations you and I had along with David Floyer, several years ago when we were talking about things like consolidation and efficiency and VM density and things like that and that's all cool. But really we get excited and John's coined this term of data infrastructure which is all about delivering massive productivity increases to organizations. And that's where this whole thing starts to get interesting, where you can actually support analytic and transactional applications in near real time. What we define as real time is before you lose the customer. Before they get distracted because our attention span is shortening. So that's where I think many people sort of underestimated the potential for data infrastructures, not just flash. I mean it's a whole new way of architecting data center infrastructure for these new emerging applications. Can you talk about that a little bit? You know, the drivers for that are the fact that we want more relevant information more quickly with more access points, right? So you've got proliferation of the devices that you're accessing it from. You want the data to be more relevant which means the back end systems, they have to stitch together information from multiple sources from larger sets. So we have some examples of a call center. It's all about how fast you can respond to customers calling in. And those call centers going from using legacy storage systems to using IO memory, they had measurable results in terms of how many more calls could be handled by per person per day. Or another classic example is in the film production, post production editing, one of our case studies, a company by the name of Prime Focus out of the UK. They've done the 3D conversion of many of the recent major blockbuster movies. And they've shown that their artists are three times more productive. They get three times more scenes edited each day because they're not having to wait for the data to load into the system. So this does ultimately boil down to human productivity. And that is what drives economic growth as well is to get more work done per unit of human input. So you guys have announced your SDK. I mean, obviously you're building on this notion of your invention of atomic rights and running the horrible storage stack, Pauline, this calls it. But there's a lot of examples, I mentioned data domain before, of just drop it in, don't change any of the existing processes. And I know you've got products that play in that market too, but to really get that productivity increase, people have to change the way in which they write applications. And I know in the past we've had this conversation, there's been some sensitivity to that, but you must have had some friction early on about that. And obviously you've made some strategic moves and acquisitions to address that and expand your TAM, but you stuck to your knitting, so to speak in those terms in your, I guess, brave in that regard. Talk about that a little bit in terms of, how you said, all right, Torpedoes, we're going there. Was it guys like Facebook who dragged into that? Was it just your vision, a combination of things? Well, you know, the world of big data is already changing by they're having, to get scale, people are giving up on data management services. The conveniences that ASAN offers, they said, forget that, it's too slow, it's too expensive in a proprietary system. So they go to a scale out architecture. So things like Cassandra, that doesn't sit on ASAN, right? MySQL, you know, all of these big properties have already moved beyond that. But what they've done is they've sacrificed the convenience of data management. And in some cases, transactional consistency with no SQL, it's like, forget consistency, it'll be eventually consistent, right? Which means you can't use it for really business critical data. No locking. Or they move to where it's not even persistent at all. It sits all in memory. So it's like, scale is the number one goal and to get that, you forego data management services, transactional consistency, even persistence altogether. Because of the business value that it draws. Because of the business value of scale. Well, now what happens with the introduction of Flash is you can offer those data management services in microseconds, millions of a second, instead of in the milliseconds that are thousands. So you're moving from thousands of a second to millions of a second. And instead of accessing those services as protocols on the far end of a straw, a network, you can access them as APIs under the application. Now the implication of this is profound. It means that this new generation of applications can offload that complexity that it used to take by having a platform underneath, which is what the SDK is. It offers the data persistence services to get transactional consistency because the media actually handles that for you. So we're talking about taking the data management services of the SAN and the persistence and transactional requirements at the app and converging them in the Flash infrastructure that goes within the server. Now we talked last year at Oracle Open World and I asked you, when did you actually ship your, because Pat Gelsinger announced the beta version of VF-Cache with the read-only, essentially, Flash card. I asked when you ship your first beta version, you said four years ago. And we had Pat Gelsinger on theCUBE at EMC World and he admitted, we're behind, John, you remember that. Yeah, yeah. So that's cool. You say, all right, pat yourself in the back. They were sort of focused on putting Flash in the array. Well, you guys were sort of changing and disrupting. But now, okay, the world that Giant has awakened. Now you got the server guy, leave an EMC, running VMworld, VMware. And there's a lot of potential there to actually start copying some of your moves, bringing in maybe some of their hardware capabilities and managing it from the server level. David Fleurer has talked many times about the place to manage it is from fast servers, not slow storage. It seems like the industry is waking up. Is that a good thing for you? Cause it's confirmation or is it scary? It really is. With EMC making a lot of Flash and noise with lightning and thunder, it's driven a lot of awareness about Fusion I.O. And being an independent company that is agnostic as to the storage system that you might use behind it and agnostic as to the server with all of the server vendors and partners, this allows our clients to come in and they can use Flash without having to be forced into using a specific storage vendor or forced into using a specific server vendor. So having the best of breed and an open platform choice is incredibly valuable in terms of driving the adoption. So guys like EMC, your biggest competitor's Oracle with Exadata, are those your biggest competitors or is this sort of host of Flash startups? Well, first off, our biggest competitor is the status quo. People using mechanical storage systems because they don't know better, right? It takes, somebody has to learn the ROI proposition to do it. Flash is too expensive. Well, they think that. Why am I gonna introduce another component? They have to recognize that if that server is able to do 10 times the workload, then you save a lot on not just the server cost but the networking cost and not just the hardware consolidation but almost more importantly, the software consolidation. And as we were talking about the business impact. The business impact. So we've worked very hard to tool up a very skilled sales team that can go and articulate the value proposition here. So our main competitor is ignorance. It's the unawareness of why this matters. And the market is so under penetrated that you can see that. 90% of the market remains unaddressed not because there's not a positive ROI but because they don't know about it. David, take your CEO hat off, public company CEO hat off for a second and kind of be kind of more of an industry guy who knows what's going on. So Pat Gelsinger and Paul Morris talking about automate pool. I mean, abstract pool and automate was a key message they talked about. I'll say that's code word for operating system. And so a lot of new things going on. Flash is a big part of that equation as an element in this new architecture. So for the folks out there that are an IT and IT professional, a data scientist, people aren't day by day knowing what's going on in the industry. How would you share with them what's happening in this market? Why this modern era, this new data infrastructure, this new OS, whatever you want to call it, we're kind of moving quickly down that path. Explain to that IT professional that data science is that analyst. If we take a step back, the key catalyst behind this is that just 10 years ago, Flash didn't exist. It was, you had technologies like it but it wasn't relevant. It's only been in this past decade that you've got a memory device that has 10 times, 100 times the density of DRAM where with a DRAM module you get maybe 16 gigabytes of capacity. You're talking about 1600 gigabytes of Flash. That didn't exist back in the formative years of operating system design. RAM was the density king. So what you have now is the answer to that. You also speed. That's where all the action happened. You load your code in RAM. Flash is able to offer you something with 100 times the density of RAM and performance thousands of times faster than disks. If that had existed back when operating systems were first being built, it would have been the centerpiece of the operating system because the operating system's basic function is to broker the data flowing into and out of the application. Remember when I first started coding we would have this jam code into 64K. Remember that to those days? Yeah, but if you had to pick a technology to put into the enterprise, it wouldn't have been Flash. Thank you, Steve Jobs, for the reason we have Flash. Isn't it ironic that the consumer electronics market drove the volume to get Flash to a low enough cost point? Not unlike the home computer market drove down the cost of processors. Microsoft and Intel grew up because of the demand for PCs for home computers and then ate back up into the server world. So you're seeing again that the consumer world is driving a new technology that ultimately eats its way back up in because one thing, and I'll have to attribute this to Mark Leslie, the CEO of Veritas. I was talking with him the other day and he said, David, this is a class example of when small meets large small always wins. And you saw that with Intel, Microsoft, versus the mainframe. You saw that with NVIDIA, ATI, versus the workstation and you're seeing it now with Flash accelerators versus a SAN. Except there's a one nuance there, which is when it comes to data and the data loss, just good enough, Microsoft was good enough, right? But just good enough is not good enough with data. And that's where the software comes in. You have to go through gymnastics to. That's why this isn't just a matter of throwing Flash chips into a server. This is a matter of integrating it using memory controller technology that makes those memory chips redundant. So failures don't compromise your data. And it's about the software to be able to replicate and share this over networks, like with ION or IOTurbine and DirectCache. Small beats large, great message, love that. David Flynn, CEO of Fusion IO, also other key themes of innovation, abstracting away complexities is always a winning strategy. And new architectures, new experiences, you guys are at the center of it. Congratulations, thanks for coming back on theCUBE. We'll be right back with our next guest right after this break, David Flynn, the CEO of Fusion IO, hot startup or hot public company, was a hot startup when we first met him on theCUBE. Congratulations on all your success. We'll be right back after this next guest. Thanks, John.