 Okay, we're back here inside theCUBE at Oracle Open World. This is SiliconANGLE.com's flagship program. We're out to the events, extract a signal from the noise. It's a special spotlight sponsored by Fusion I.O. One of the great supporters of theCUBE that allow us to bring our great independent coverage to wall-to-wall exclusive interviews with thought leaders, experts, CEOs, alpha geeks, and we love it, and we want to thank Fusion I.O. for that. I'm John Furrier, the founder of SiliconANGLE.com. I'm joined by my co-host. I'm Dave Vellante, wikibon.org, and we're here with Brian Bulkowski, who's the founder and CTO of Aerospike, new type of database company. Brian, welcome to theCUBE. Hi, Dave, glad to be here. So, through all the action is John. Yeah, we all know each other around Silicon Valley. He's making it happen. So first, give us the update on the company, right? So Citrus Leap was the old name. Aerospike's the new name for the folks out there. Introduce them to the new company name and update on what's happening with you guys. Absolutely, so we founded the company four years ago as one of the fastest databases in the world. We focus on commodity hardware and especially on flash storage. It's a special focus of ours realizing that companies like Oracle had great technology with rotational drive and the world was going to switch to flash and DRAM and the time had come for a new database. So we've been selling our database primarily in the fastest, most scalable platforms in the world, which are mostly advertising platforms. These are folks who are doing 10 to 20 billion impressions per month. We've had no downtime with any of our customers and we can usually solve with flash storage, multi-terabyte systems, billions and billions of rows, often for user tracking and user information, personalization of advertising and experiences. So we've gone last year from a great revenue year into this year, more sales, more customers. So I got to say that I've known these guys for a couple of years since they've been working on a great company. Brian is being humble, he knows what he's doing and saw the stuff flashed up early. So I want to say congratulations, great to have you on theCUBE. So I want to ask you, what did you guys see early on that you wanted to tackle? So when you started the company citrus leaf at the time and now Aerospike, what was the big thing you wanted to focus in on for the company? Sure, John. So what I saw was companies that were being strangled by scale. I was at a Kleiner-backed company and they were trying to cover 5% of all the internet traffic in the world. They were trying to build dynamic recommendations on websites, great dream. And they were using Sharded MySQL, which was the style of scalability at the time. And they found that to be a really difficult chore. They had outages, the moments when they needed to scale most and needed to absorb the most traffic. Those were the moments that they were down. And I said, you know what? There's got to be a way for the highest levels of scale because the entire world is going to be moving to more and more scale, more and more data. And big data, Hadoop was already upon us, analytics, business intelligence. That was a train that was already had a lot of velocity and a lot of momentum. But transactions, fast transactions, that was an area where there was going to be a lot of growth in the future. And especially with Flash, frankly. I said, well, it's all about multicore and distributed systems, but how that works together with Flash is what the next revolution and databases will be. Well, it turns out it absolutely happened. So let's take us to today in terms of your current situation. How big is the company and what's the big product news? So we've had Citrusleaf, that's the old name. We've had Aerospike2o out for nearly two and a half years in production with about several dozen customers. Again, a few at the highest levels of scale in the advertising industry and also small customers. So folks with really big dreams who want to grow and don't want to fool around with some open source solutions. So we've been expanding, continuing to conquer within the advertising world. Companies such as Exolate, great customer of ours that we announced a little while ago. But especially data management platforms, the guys who are, okay, maybe it's a little creepy, but they're taking all the data about everyone on the internet and they're analyzing that and then feeding back out and syndicating all of that. So every single person all the time, millions of transactions per second, and they have to be up all the time. We've never had an outing. Brian, we've heard a lot this week. It's only Monday. We've heard a lot from Oracle about Flash. I always joke, like they invented it. So, but you had said you founded the company because you saw that the sort of traditional systems, disk-based systems weren't going to cut it. So as a observer, you listen to Oracle's messaging. Sounds like they've hopped right in the bandwagon. What's different than what you do? We've built our system from the ground up to be Flash-oriented. We work with main memory and we also work with Flash memory and we don't really support rotational disk at all anymore. We use it as a backup store, as a warehouse-style technology. That's where we keep things like transaction logs in the database, but we don't serve from it anymore. And that's been true for years in our product and that's why we can achieve things like recent blog posts, one million transactions per second per server out of main memory DRAM and that's for a several thousand dollar server, nothing fancy. What we showed with Fusion I.O. with their new native interface at the demo conference a few months ago was 400,000 transactions per second on a single fairly modest server with Fusion I.O. hardware and their special native interfaces. But this isn't a clustered environment. We've had, I've had customers come to me, guys in doing something new and exciting. They say, my business only works if I can do two million transactions a second, but I need to do it reliably and I need to do it with only a few servers. Can you help? Because there's no other way to do it. And I'll be interested to see what some of Oracle's claims are, but we've had that kind of technology for a few years now. And we've had it in production, it's a hardened solution. So you're essentially saying, if it's on spinning disk, you don't want to access it. Not really. This is going to be too slow. No. So every, all active data will be on flash and that's really the fundamental architecture. Exactly. And the main change is when you think of users, they behave irrationally. They're hard to predict. That's the whole point of users. They do all kinds of things. So you can't use caching technologies. You can't assume that the most recent users will be the most recent users a few minutes from now. And so, and the analytics you do, it might not be on any particular pattern. So the typical tiered storage approaches, frankly, have a lot of trouble. So we said a better technology, looking at flash and looking at distributed systems, is one where you just have one database, no cache. So our system is as fast as cache you put in front of a database, but you can build 10, 20, 100 terabyte systems with flash storage, cost-effective way, reduce the complexity of your architecture. No database caches. What a pleasant world. Can you explain why that's such a much more pleasant world and what the drawbacks of a database cache are for our audience? Sure, cache systems, you have to warm them up. You don't know what's in them. When a cache, you have to arrange consistency between them. When you have a cache tier, well, usually you think of a cache tier and you say, well, that's okay if that's unreliable. It's okay if that goes down. Well, that's not true. Because if you didn't need it, you wouldn't have put it there in the first place. So instead of having a cache tier that you have to manage and then a database tier that you have to manage separately, just do it once, do it fast, and do it cost-effectively. So you're taking advantage of things like atomic rights and in the future roadmap, or how does that all work? So I sat down with, actually in the very early days of the company with the then CTO of Fusion IO, David Flynn, who I knew from previous company. And I said, David, I know flash storage is going to change the world databases, but I don't really know it. We had one of those 15 minute conversations where he took me through the entire internal architecture of Fusion IO, how the whole thing worked. And I said, hey, wait a minute, you have a key value store. You actually have an index database inside every single flash card. There's a little database in there. Well, heck, I got a database too. I'm doing a database and you're doing a database. That's dumb. One of the best ways to get speed and efficiency is to collapse layers sitting right next to each other. So the inside out of that was that the way that disk is optimized, storage is accessed is through sectors and blocks. There's no sectors and blocks here, right? There's chips and essentially a key value interface. So I said, well, hey, what if some number of years from now we all get together and we build an entirely new kernel interface for storage based on keys and values? And we all said, ha, ha, ha, well that's great. Let's go back to our lunch. And then they went public. And we got our fund and we went and rolled out to some great, very high-velocity customers. And time came for us to sit down when they were doing their most recent hardware rev, the Duo Drive 2s and the two solutions, and said, hey, let's build that dream. Let's actually build a key value interface for flash and let's tie it directly into the database. What new levels of efficiency can we get both in terms of, as a database developer, I can build a more atomic, more consistent solution at a higher level of performance. So there's a whole trend that Fusion IO is doing a great job with with these atomic memory operations. But we actually took a different direction. I think Fusion IO is really to be commended for this thought leadership of not just doing it one way or doing something new, but really working with multiple companies in terms of bringing out some of the efficiencies and really driving some of the benefits that can be done out of flash. Because, you know, as I said to a VC a few years ago, I said, well, you don't support rotational disk. And I said, yeah, we don't support paper tape either. So can you talk about, so that's great. Appreciate the explanation. What does that mean from a business value standpoint? Can you sort of net it out? What are your customers seeing? Faster, cheaper, faster, cheaper. Well, more reliable too, but faster, cheaper. Can you quantify that in any way? I mean, in terms of bottom line productivity? Well, what it's about in terms of a system architecture, like I said, is simplifying and doing more that you couldn't do before. For example, I was talking to a company down in the Wall Street area about particular calculations that they do that also are very random. So this particular fairly large company, I said, look, you've got every database in the world. You've got Oracle, you've got SideBase, you've got all the old IMS stuff, you've got Vertica, you've got everything. Why would you come to us and start talking about a new database? And the answer was there are random access patterns in their calculations, what's called what-if scenarios, where you don't know what's gonna happen. You need to game out various different strategies. That cannot be done with rotational disc or cash optimization strategies. You have to have a very random, very fast solution. So what that means in terms of the bottom line of a business is let's take our advertising customers. Very cutthroat business, it's all about the amount of the pennies that you shave on every transaction with a, say, 25 server cluster, we're able to serve, say, 25% of the advertising load of North America. So if you can have that kind of bottom line business impact, why won't this, or will this, technology go mainstream? It would absolutely will. It will, that's why it's kind of another round of round. What are you gonna ask me? No, it's not gonna go mainstream. Well, I don't know, I mean. It's early. It could be a nice niche. It could be a nice fringe. But what you're describing is not fringe, it's the future of- It's such a man. Well, no, he's saying it's random now because he got his legacy installed base. And what's happening is the random use case he's got is going to be the preferred architecture. Because that's going to be the color the cost drives down. That's what basically you're saying. Because right now- I talked to several companies recently that have men cashed here. So in memory DRAM, men cash tiers with, and they have four to five terabytes of DRAM. So they bought a lot of servers just in order to have a caching tier to make their database go fast. And we walk in and they say, well, we've got all these problems, I've got this whole ops team, we're supporting this. They're apps we can't build because we have to keep expanding this tier and I can't put in another, I can't double the size of that tier cost effectively. But what if I was using Flash and what if I didn't have to keep an entire database behind that? So we believe, I believe that we'll see many more applications using high velocity, fast data. And we'll see new applications because of it, whether it's mobile, where people are doing new things constantly and moving around, whether it's in the advertising space, the gaming space, we're talking about some gaming providers who need to, they need to cut new games, right? And they need to push the edge in terms of especially social, where sharding doesn't work. You can't just have some people over here and some people over here. Now everyone has to be linked together. There's a variety of places now where folks are doing things that they could never do before. I mean, you guys had some serious technical accomplishments under your belt and with that, I wanna ask you, did you lose your headset? Kian, can you help us out? So Dave, let's talk about the technology, obviously scaling. So I wanted to ask him Oracle. I mean, that's some technical accomplishments we have there. In Oracle? No, Oracle is promoting massive scale and he's on the cutting edge. Well, I think the point, Brian, you were making earlier, is that essentially what Oracle's doing is saying they've taken their existing architecture and jam flash into it. And we saw that originally with storage array companies. So I opened up, put this flash into this storage array and worked okay. But it wasn't game-changing. It wasn't an order of magnitude or two orders of magnitude performance improvement. Exactly. So, I mean, part of Fusion IO's business, a great business is you take the Fusion IO card in and whether it be MySQL or Standard Oracle or Rack, and hey, your database goes two to three times faster. Just magic pixie dust and a little bit of money and you just have to bottom line out whether that's worth it to you. Great, that's a factor of two, a factor of three, maybe a factor of four. What happens when you need to go 10 times faster, 100 times faster? What happens when you have to do multiple database lookups for every single ad served on the entire internet all out of just a few servers? So we put the fourth to premise that it's going to go mainstream, but there's got to be a tipping point, right? Because you're essentially talking about changing the way in which you write applications or maybe rewriting existing applications, right? So that doesn't happen overnight. How long do you think it's going to take? I think we're probably still two to three years minimum as it's starting to be accepted. So the leading edge is, I believe, gaming and also advertising. Those are the folks who have the highest levels of load and also in very vibrant competitive industries. Probably it will take in the order of five years for people like insurance companies who have to do fraud analysis and the companies that serve them to really start making even deeper and deeper use of this kind of technology. Brian, I wanted to ask you to comment on Oracle's statements yesterday in the keynote. Larry Ellison was talking about all this performance with Exadata. Larry who? Ha ha ha ha. So once you're coming, you're in the cutting edge of really a work that's going to be the future. And he's trying to co-opt the work that everyone's kind of doing and he's overreaching. And I think there was a tweet that said you get four times the performance with Fusion I.O. Someone tweet, I think it might have been someone from Fusion I.O. So what's your comments on Larry's current positioning on the tech involved? Well, I didn't see the keynote, I got to say. What we really strive for is those 10X and 100X of performance. And that does require new interfaces. And I got to say, I do like Oracle's MySQL Cluster product. We've seen it head-to-head competitively. We beat it every single time. But it's actually a very good product and I think does show Oracle validating the NoSQL space and realizing that there's something to be had there. It's a great space, you have to do something other than throw some hardware at it. That's great for some old mainstream apps. But an Oracle Confluence for a long time has been an in-memory data store with user-defined functions that can be used for a large variety of applications. Of course, it's main memory and they've, as far as I know, never done a flash integration. Proved me wrong, but Oracle actually has, besides its main core database, a large history between times 10 coherence and MySQL Cluster with the NVV engine in NoSQL. So I welcome them to the party. Final questions, we've got a break on time. What's your plans for next year? What's the roadmap for you guys looking for over the next year? Sure, so we've done two things that we're very excited about. One is last month we introduced Aerospike Community Edition. So that's a two-server version, so folks can experience the power of flash storage as well as the interfaces we propose, the speed and also the reliability and two-server clusters. So we invite folks to be able to start their companies, start their businesses with that level of scale. We're starting to work with essentially high-speed analytics and faster queries. So we believe that there is space within the market for queries that are still happening in real time. They're on the front edge of the database or they are analytics that have to be very, very fast, such as financial services analytics where you're changing second by second the algorithms that high-frequency traders use. The very highest speed algorithms are at the low level, the chip level, but every few seconds there are analytics that are done to switch the course of the boat. There's a number of areas in there that we find fascinating that go beyond sort of the simpler key value use case and we're already in alpha looking for customers in that area pushing out some new technology. Brian Volkowskis, the CTO founder of Aerospike, part of a new generation company, congratulations. We're seeing companies like Fusion I.O.Go Public setting the agenda for the new infrastructure. You're part of that wave, congratulations. This is SiliconANGLE.com's theCUBE. We'll be right back after the short break. Thanks, John.