 SiliconANGLE's exclusive coverage of Oracle Open World 2012, this is theCUBE, our flagship program. We go out to the events to extract the signal from the noise. I'm John Furrier, the founder of SiliconANGLE. Joy with my co-host. I'm Dave Vellante of Wikibon.org. We're here with our good friend Gary Ornstein, many-time CUBE guest. Gary, welcome back. Thanks for having us, gentlemen. Appreciate it, great to be here. You had a rapid-fire segment. We only had a short time at VMworlds. You were on theCUBE at a short time, but it was short and sweet. You really knocked it out of the park. Well, short is a nice short and sweet, right? Everybody likes that. So, Oracle promoting real performance here. So, give us the two cents on your opinion of Oracle's whole Flash vision, because you guys at Fusion are crushing it and leading the market. Yeah, well, I think the good news is, everybody's recognized that the classic infrastructure that we've had previously, largely based on mechanical disk drive, isn't enough to cut the mustard, so to speak, with today's data workloads. And you've seen everybody, both Larry Ellison, and then I wasn't at the keynote this morning, but I understand Joe Tucci was talking a lot about Flash memory. It really is, to some extent, a broad recognition that this is an imperative for dealing with today's workloads. And it's just great to be at a show like Oracle OpenWorld, where we see so many customers who are pushing the boundaries of big data, small data, every size data, and various configurations, and looking for ways to do more processing, more quickly, but also do it with less infrastructure. So, that's a big theme, is we want to increase the performance, but we also want to shrink the data center footprint. I mean, nobody wants a bigger data center, but they want a smaller data center. You know, you must love it when companies like that, Ellison, Tucci, you know, big exec stand up, and essentially validate what you're doing. It was interesting to hear, I don't know if it was either Tucci or Jeremy Burton, talked about, I think it was Jeremy Burton. In 2008, they shipped the first enterprise Flash Drive, and I tweeted, yeah, right around the same time that Fusion IO was going to market with its, you know, PCI. I saw your tweet as I was coming in this morning. I say that only because at the time that EMC put that enterprise Flash Drive in, we said, wow, this is going to change everything. EMC landed a haymaker. Yeah, yeah, Fusion IO, that's kind of interesting. Then we learned more, that's really interesting, but it's somewhat nuanced. But you guys are really changing the way in which people think about Flash, but it's hard to get people to change. You know, it's different, and in any technology market, enterprise, even some of the web market, which the web market does move a little bit more quickly, but not overnight, but the enterprise market certainly takes time. And I think we're at a stage now where people, there's enough success stories and enough customer momentum and enough people have deployed Flash memory in innovative and successful ways that, you know, we're just seeing this continue to grow. I think the interesting thing now is that, you know, customers can have a choice of servers they want to buy. They can have a choice of the deployment method. You know, we can, we have customers who deploy Flash in the server as local storage, who benefit for that maximum amount of acceleration where the data and the CPU are right next to each other for the lowest latency transactions, the highest amount of throughput. We have customers that use Flash memory in conjunction with their existing storage environment to do caching so they can retain the best of the data management tier, combined with the best of a performance tier. And then more recently, we introduced the Ion Data Accelerator software, which allows customers to take Flash memory, put it in a server, but then share that as a shared resource with other servers. And so now, any which way that customers want to consume Flash memory, they have that at their fingertips, which is just to get another accelerant to the market in general. Gary, so a lot of, as everyone's racing up to catch up the Fusion IO because you guys pioneering the Flash area, obviously EMC is moving, like pedaling as fast as they can. So the demos on stage, Jeremy Bird is going to come on later. He said, I'm Dak here, Oracle with their moves. How do you guys look at that race? Are they catching up? And what are you guys doing right now to keep that distance ahead of the pace? Well, we think we're ahead in terms of the product portfolio. One of the important things is we really believe that customers should have a choice of the hardware that they deploy in their environment. And we've worked very hard with our OEM partners, folks like HP, Dell, IBM, Cisco, Supermicro, NetApp, we recently added as a partner. And so we want customers to be able to deploy high performance configurations, but with their choice of their favorite server. And so that isn't always the case with some other options in the market. Often customers will have to stick with one single solution. Sometimes it's proprietary and doesn't provide a choice of options. So we see it as an important element of our company and our product portfolio to give customers that choice. Choose the server that you want to deploy. Add FusionIO memory. Turn that server into the personality for the workload it's addressing. Should it be local storage? Should it be used? Flash memory for caching? Should it be flash memory as a shared resource that you can go across your network? But always have that flexibility. So we had Brian Wilkowski on yesterday from Aerospyke, formerly Citrusleaf. Really amazing company. I tweeted out this morning, hot startup to watch. These guys certainly have an awesome path, but we had a lot of people talking to me last night saying, hey, what was he talking about that FusionIO connection? He had talked about it, he had a meeting with Mr. Flynn and they talked about the two databases. How does that, those guys fit into Fusion and what was that dynamic that he was referring to? Let me abstract that question a little bit to talk about the software development kit activities in general and I think specifically the example that Brian was referring to. So when you're managing flash memory, you have to keep track of the data and the placement onto the physical chips. There's an inherent organization that you have to do. The technical term for that is called the flash translation layer and so we've been doing that in our products for a long time. Essentially customers have a piece of data that they want to place and we keep track of placing that on the flash memory product. Now that is essentially a data translation layer, a little bit of a database if you will and so if a piece of software on top of an IO drive is also doing that same kind of I have data and it wants to go somewhere and then you have two mapping so to speak, two organizational constructs and what we're doing with efforts with companies like Aerospike and our software development kit is trying to collapse those layers so that we're operating in a more integrated fashion and so instead of having two steps to go from what the application wants to do down to the physical media, maybe we can shrink that to just one step. Now what's the benefit for doing something like that? Well, it means that the application developers can write less code. It means that they can rely on the Fusion IO products to do a considerable amount of the heavy lifting so to speak so they can be more application centric and we can be more centric in providing them some of the underlying services. It can increase the performance and again it fits with our whole strategy of trying to make things more simple by removing some of the layers just like in introducing our product we removed some of the antiquated storage protocols. We don't need those layers anymore with flash memory. We want to treat that like a memory, not like a disk drive. Same thing comes when we're working with companies like Aerospike doing these new data stores, let them focus at the database tier and we can provide some of those underlying services. So, hold on, I got to ask one more question. So is there confusion between Fusion, Oracle Fusion and Fusion IO? I mean, it's a legitimate question because Oracle Fusion is their middleware, right? So, you know, it's a different product. There is Oracle Fusion middleware where Fusion IO but if somebody gets confused. I think you might get the better end of that deal. That's all right. I mean, we won't complain about that. But it's a good point. I'm glad you noted it. So you're talking about how you've essentially removed the need to go through what Pauli Nist of Intel calls the horrible storage stack. We love that term. So you've created that innovation but there's another piece of that too which is that application developers have to take advantage of that. Correct. And that's where your SDK comes in because you can update us on the traction that you're getting there. I mean, it really is a land grab for the developer community to actually start running then. To the extent that it happens, you know, it's just a bigger rocket ship. Yes. But it's, you can't just throw it out there and say, okay guys, go. Yeah, it takes work. And we were fortunate yesterday to hear from Brian Bukowski at Citrus Leaf and the folks from Couch Base who we've also talked about publicly. There are a number of others as well. And for folks who are interested in the program and the developer activities around the software development kit, they can go to developer.fusionio.com as a starting point. But we are seeing a tremendous amount of interest as people sort of wipe the slate clean. Everybody knows that there's opportunities to take advantage of regarding the use of flash memory, particularly in using it more like memory than using it like a disk drive. And so we're working through a number of different areas there. Another area that we've spent a considerable time with is something called Atomic Rights in conjunction with MySQL. And recently we had some news from Percona because Percona's been doing a lot of work here and trying to include that in the software that they deliver as well from a MySQL perspective. Atomic Rights, very simply, you have multiple pieces of data that you want to persist simultaneously, altogether or none at all, a way to keep your database tightly cohesive and coherent. And so with disk drives, the database would have to write twice to keep everything in sync and in case there was an issue. And just the disk drives aren't capable of doing all this stuff, but with flash memory, we can do an Atomic Right all at once. All those pieces of data can be persisted simultaneously and guaranteed. So the benefit of that is there's less code that's needed at the database level. There's faster performance because of that. And then also in the case with flash memory, we're writing less to the drive. And so that's going to extend the life of the drive. So if I may, prior to the Sun acquisition, that would have been music to Oracle's ears. They would have hopped on this at an order of magnitude, better performance, make life easier for database people. So where is Oracle in this? They're not hopping on it, right? Probably the same true for some others. Well, I can't comment on Oracle's strategy. You'll maybe have Larry coming on the show later. Yeah, well, we have one Larry Johnson. He's coming after his keynote. Talk to you after the keynote. Right after the keynote. We'll ask him. But we are seeing a lot of interest from the people who use Oracle products. We are doing work with the MySQL community, which is still a vibrant and important part of this whole world. So there's still a tremendous amount of opportunity. Yeah, okay. So my question is on the market. So one year later, last year, Oracle Open World, we are here. Describe to the folks what's happened in one year from your perspective. You're out in the trenches. Obviously the market's been growing very, very fast. What dynamics, what kind of notable trends can you share with the folks out there of things that have happened in the industry that they might not have read about or be aware of? Yeah. You know, I don't know if there's any major singular dramatic change. I think there's a couple of things that are becoming more well understood, as we mentioned at the beginning of our discussion, the just awareness of flash memory in general. You find people who are less, you know, who are more savvy about what's going on, who know that there's various options in place. You know, I think on the Oracle side, we continue to see people who will take advantage of built-in tools to the Oracle database, software stack, things like ASM, the automated storage manager, so they can take advantage of flash memory and organize volumes just right within the database using tools built into Oracle like Data Guard for replication that, again, is just built in. So it's very simple to do very, it's very simple to do sophisticated configurations with Oracle and a variety of deployment methods and Fusion IO memory. And a lot of it is just right out of the box. So I got to ask a startup question because you've been in the startup, you've been part of, you know, the whole startup scene especially in Silicon Valley here. We were talking earlier about networking with Lee Doyle, XIDC analyst about networking, how it takes a lot of money to start a company. Well, storage now, same kind of thing where you know, do a startup, you cost a lot of dough, but flash is creating a lot of opportunities. So what's your observation around startups that are doing well in this new world? Because there's the old way and there's new way of storage. You guys are doing deals with everyone, you're seeing a lot of action. If you can just kind of describe things that you've seen people doing that are very successful as a startup, whether it's product, their approach, their architecture. Could you share some perspective? One thing I want to mention about the storage market which is a place where I've spent a lot of time over my years is things don't happen overnight in the storage market. And you know, it's important to remember and what I'll call at least in my tenure in the industry wave 1.0 of the storage startups, these array companies that all were birthed and acquired within similar time zones. So we had time range, propellant, That's called a bubble. Equalogic, three-par, Isilon, Data Domain. Have them are, you can do it. They were all, you know, all those companies, there was this wave of acquisition but those companies had spent 10 years in the trenches and storage successes do not happen overnight. Now I think we're in a similar scenario where we're seeing a lot of folks pop up on the radar who are doing new types of storage systems. I think the interesting ones are those who have said, I'm not going to treat flash memory just as a disk drive. I recognize that this is a different medium that it has many positive characteristics that are totally unlike disk drives. And I want to create a system that will take advantage of that. And we do a lot of work with storage system companies like Nutanix and NextGen and Kaminaria. Nutanix is going to be on later today. Who integrate FusionIO products into very successful systems. And so I think the secrets to success there are going to be those who recognize that there is an opportunity for a twist on the conventional architecture. But again, it won't happen overnight. It does take time and we're seeing that already some of the successful ones break away from the path. So you've got props, I see. Yeah, well, this is an example of our IO drive too. This is a life size, but it's a thin one. That's the more interesting angle. Yeah, it's a cardboard cut out but we're giving these away at the booth. You know, it's important to hold it and recognize that something so compact and so small can deliver the performance of what formerly would be required in a full size rack of disk drives. I mean, one memory device can deliver the power of thousands of disk drives. And so sometimes, you know, when you're out of show with people, it's important to say, hey, you know, this is all it takes. Fantastic. So now this is a new product cycle for you guys, right? Yes, well, we introduced that late last year and that's going great. And then, you know, more recently, I mentioned earlier we're working on a lot of software products, both caching software and then this new Ion Data Accelerator software for the shared storage model. So do you consider yourself a software company? Sure, yeah. I mean, they're software all the way throughout. I mean, VSL, which is going back to that, how do you place data and organize it logically to physically on the device? Our virtual storage layer, which acts as that flash translation layer, has been at the core of Fusion IO products since we started shipping products in 2008. And so that's a very battle hardened piece of software. And then we're now coupling more software capabilities on top of that primarily to give customers ease of deployment. Again, use the product in server and benefit from the local storage access and the speed and the low latency. Use caching to integrate with your existing storage systems. A lot of work going on with NetApp from the caching perspective as one example. Their products are great fit with ours because they also use a log-structured file system. And so, you know, when you offload some of the read activity, fantastic performance from those systems. Again, or the Ion model of use it as a server based flash sand, so to speak, that you can be shared across different products. So when you look back, I don't know, today, let's say 10 or 15 years ago, you had fast storage, fast spinning disk, which is sort of an oxymoron, and you had slow spinning disk and you had tape. And I want your opinion as to when you look forward, 10, 15 years from now, you look back. But so, we've seen tape just sort of disappear as a primary backup medium. That's gone, right? I mean, it's mostly disk based backup now. What do you see is that hierarchy? What's it going to look like? Is all active data going to be on flash? Is all data going to be on flash? Is disk just going to be for backup and data protection? What's your view? Well, one thing I've learned is, you never want to predict the death of any media. And media's come and go. And that's okay. You know, there's nothing wrong with that. And you know, we don't necessarily need to get sensationalist about declaring the death of X, Y, and Z. Even newspapers don't die. Even they don't die. But I do think we are now seeing a well-recognized scene of separating the performance tier from the capacity tier. But here's an interesting twist. In this race for the performance tier and the race to process more data, and the race to serve more users, and the race to capture every last click in some kind of big data Hadoop configuration because we don't know when we might need to analyze something out of it. In the race for all of that, we're just generating more and more data. And so there is a symbiotic relationship between the performance tier capturing and measuring all of the activity and the capacity tier needing to retain that. Whether that be in a classic storage system from the likes of a company like NetApp or in a new fangled system using HDFS or similar. So that I think is on the horizon. And I want to tell you about what I see with it. It's really what our customers see and the customers that we see who are deploying Fusion.io in these real-time environments, when they're processing their transactions and their application, they don't want to leave the motherboard, right? That's what they say. When I'm processing, when I'm doing my business and my CPUs and my application are trying to serve my users, I don't want that CPU to have to leave the motherboard. And so they're moving to this scenario where they want to be able to process everything at the application level in memory, in real-time, but then also have the retention and the data management. And that will most likely be a heavily dis-based infrastructure. Okay, yeah, great to have you on theCUBE again. You guys are doing great. Fusion.io leading the charge. Going back to VMworld 2010 with you guys private and then go public and now continue to push the envelope, bring in the software development, get in. Truly making it addressable and scalable. Congratulations. Okay, we'll be back with theCUBE right after the short break.