 Live from the Moscone Convention Center in San Francisco, California, it's the queue at Oracle Open World 2014. Brought to you by headline sponsor Q-Logic with support from HGST. violin memory and mark logic. And now here are your hosts, Dave Vellante and Jeff Kelly. Hi everybody, we're back at Oracle Open World 2014. We're here live at Moscone South inside the Q-Logic booth year five for us at Oracle Open World. Jeff Kelly is with me. Jeff, great to see you. Really happy to have you here. We're going to roll into the afternoon with your perspectives. Good friend, Brian Bulkowski is here. He's the CTO and one of the founders of Aerospyke. Long time Cube guest and good friend of the Cube, Brian. Welcome back, good to see you again. Oh, thanks a lot. It's always good to be on the Cube. Yeah, so Oracle Open World, Oracle, let's start with Oracle and transition. Let's just start there. I often sort of say tongue in cheek. Oracle will be late to a trend and then kind of act like they invented it and we've seen it now with the excellent Oracle marketing with 12C and one click to the cloud. What's really going on in the Oracle world? Well, I think everyone's realized, hopefully everyone's realized that the days of selling software are numbered. Right, the idea that you're going to actually buy a piece of software and install it somewhere, the writing's on the wall. No developer, no software engineer really wants to install software anymore, they simply want it to be there. And Amazon has paved the way. Azure, part of that, all the Paz guys, what Salesforce did with Heroku, those are all paths on the way to I don't ever want to install software, I want to have an environment, I want to trust that environment, I want that environment to work. Well, Salesforce kind of had that vision in 1999, right? No software in here in 2014. Now, it seems like from a business model standpoint, Oracle's genuinely accepting that, cloud business is growing, Stafford talks on the quarterly conference calls about how committed they are to wrapping hardware and software together, I.I. infrastructure as a service, pass, database as a service, software as a service, the whole nut. Can the new stuff in your opinion grow fast enough to offset the decline in the old stuff? I think it's going to be hard to get anyone to switch over to Oracle's cloud, because data has gravity. Once you have your data in one place, shifting it, going back and forth with that data, it's very difficult and this is where Amazon and EC2 has really stolen the march on so much of the world. People's data is in Amazon, getting that stuff out of Amazon is going to be a trouble, right? So the people who are already in Azure, et cetera. So the tweet that I saw was the damn Google, sorry, that Oracle is starting this with like 20,000, 30,000 machines for their cloud. That's only going to be the die-hards, right? The people who are, I'm going to use Oracle no matter what, compared to the millions of machines that EC2 has and the millions of machines that Google has as part of Google Compute Engine. So I think it's going to be a little tough other than the die-hard guys. Well, I think, you know, that's bringing up a good point. I mean, from a strategy standpoint, they can get the die-hards into the cloud, but the scale is not, and Oracle, Larry said last night, we're going to compete on price with Google and Amazon and Microsoft. How do they do that without having that massive scale? I don't think price is the issue. I think he threw that away, it's a great giveaway. I mean, yeah, we'll compete on price, but developers want scale, they want ease of scaling, they want APIs that they already know and love. I think this is the benefit of what Pivotal's done with OpenStack and OpenClabbing. OpenStack allows you to say, hey, you get your same APIs. I'm a little confused as why Oracle wouldn't line up behind that effort, that really being the anti-Amazon move at this point, as opposed to setting up their own system. Well, they're kind of taking an Amazon-like strategy and just building it out through the full red stack and saying, all right, we're going to be the red stack, do everything Amazon does, plus, and forget that OpenClab. Which makes sense, because they have their own application environment, all of the, they do have, in fact, a full stack suit to nuts. Whether that's enough, I think that's the question. Well, they definitely see the world differently. Well, let's talk about Aerospike. Give us the update on Aerospike. It's been a while since you and I talked and I'd love to hear what's new with you guys. So the biggest thing in Aerospike is that we open sourced all of our software. So the fact that the Aerospike database is now open source, we would like anyone who's considering Redis or one of the older in-memory solutions to really look at Aerospike. We've been trusted for in-memory deployments throughout ad tech, marketing tech, with a great suite of customers. So five plus years in deployments with very high availability systems, extraordinary performance, and now with open source, we think any developer really should just reach for a proven system in in-memory. So why the decision to open source and why now? So in our original idea of the business, we wanted to make sure that we had a great beach head and a great piece of technology. So for our initial investors, we said we're going to keep it focused. As a small company, we want to land into a market. We want to hit user data within advertising and make a great product for those guys. And then later stage, you say, okay, now we want adoption and we're starting to see that. So the guys at DataStacks are now saying, this is the year of NoSQL in the enterprise. We see that at Aerospike. We see telecom companies doing QOS based routing for which they need an in-memory database. That has to be right on the front line. We're seeing financial services companies trying to respond to mobile and the demands that that has, there's this core database, position of record database has to be in memory and has to be very high reliability. We're seeing those use cases pop up. We're seeing stacks that used to be a storage device, a database and then a cache. We're seeing all of that slimmed down to just the database. Wouldn't you rather have a database that was fast enough and didn't require a storage layer, didn't require a cache in front of it? All of those, the enterprise seems to be responding to that. And so for broad adoption, we believe open source is the answer. This is a flattening of that pancake stack. Jeff, what do you take, what's your take on all this? And we mean memory database has been around ever since I've been in the business and all of a sudden things are exploding again. What's your take? Sure, well, I mean, the price of memory has come down significantly. You've got flashes enabling some similar capabilities but at a lower price point. So I think it's certainly an area where there's a lot of interest around putting those applications right at the front line. So you're actually making real time recommendations, real time insights that are enabling users to actually get more from the application that they're interacting with. And we were talking Brian a little bit just before we went on air about, you mentioned finally this is the year of kind of real time. Talk a little bit about that both in terms of the use cases where our spike excels as well as some of the things we're seeing on the Hadoop side with Spark and some other things. So the big use case that I'm starting to see coming out of the Spark community, by the way, I'm works as excited about Spark as anyone else. I think it's a really interesting architecture. And what it seems to be used for from people I talked to is looking for things like patterns using machine learning in real time. So instead of having to sweep through all of your data with Hadoop, you might have 10s of petabytes that might take six hours. What if I can see those patterns right now? What if I can see an interruption in my cloud services? What if I can see out of my server logs a certain class of users just starting to hit right now? That's what all that stuff is great for. Now, air spikes role in that. First of all, we may be able to improve Spark simply because a shared database as opposed to in memory on the Java stack may provide some interesting computation. And so we're working with the guys with Databricks on that and try to figure out our own solution. That's great. But the real point is you want to make all of these insights part of your online experience. You want to offer a deal. You want to offer a cross-sell upsell. You want to offer a recommendation. And that's in the web experience. It can't be batched. You want to line it up right now. You want to see the insight with a system like Spark and then you want to feed it into the online experience. That's what gives people the feeling that you're listening to them and personalizing everything as part of the web experience. So take us back a little bit. Give us a little bit of a history lesson because we know, of course, that kind of personalized advertising has been around for a while, but it hasn't always been this real time. It's been more batch-oriented. Talk a little bit of, walk us through the history of how this has evolved a little bit. Sure, so if you remember around year 2000, the idea of scaling out the internet was let's get a whole bunch of MySQL. Let's shard it out. Let's put a cache in front of it. Let's buy ourselves a storage layer and a sand behind it. And that sort of took us through to, I would say, around 2007, 2008. The large internet companies like Google never even went that direction, really, is the guys trying to build internet experiences. So in 2008, something very interesting happened in technology, which was the personalization of advertising. Instead of buying an ad on a website or buying an ad on a magazine page, people said, hey, let's try to buy an audience. Let's try to buy a person. And then they said, well, wait a minute, we have to price this. So if I can get your eyeballs right now, maybe on some wacky site that you go to, driving cars in the mud, who knows what. But I'm buying you. How do I price the opportunity to get an ad in front of this particular person at this particular time? And a couple of smart guys said, hey, let's run auctions. Auctions are the only fair way of pricing the millions per second of ads that can be delivered at any moment. So companies like App Nexus out of New York, the trade desk on the West Coast, Google internally, and also Facebook now as part of Facebook Exchange said, let's run auctions. So this is very interesting because it provides a counterpoint to capital markets. Capital markets, when you trade and you buy and sell, you say, I'm willing to buy, I'm willing to sell. And when those two points overlap, then you've got to sell. Instead, you say, I want to sell, let's run an auction. And let's do that today. That's happening in advertising three million times every second, with a hundred companies bidding on every single advertisement. It is a massive piece of technology to price in real time every single ad opportunity. Facebook does it as part of Facebook Exchange, Google's ads flow out, all of these systems. Well, so the players you mentioned are obviously huge technology companies, they're big web-based companies. So they have the wherewithal and obviously the need to do that. So what about the rest of the world that could benefit from that kind of capability? How do they get access to that kind of technology, those kind of capabilities? Well, this is one of the reasons we formed Aerospike. We wanted to bring the kind of technology that was inside Google, allowing them to scale out into other companies. So Oracle, we're here to open worlds. I mean, times 10 was a great memory database. And yet the people I talked to using times 10 say, oh, but it doesn't scale. We know it doesn't scale. And I'm not a times 10 user, I can't speak to that eloquently. But what we did at Aerospike was we said, let's use the in-memory technology that companies like Google are using, like Yahoo are using internally on their own projects. Let's bring that to a wider market and allow people to compete with these kinds of in-memory solutions. Now, beyond, obviously you've got to have the technology component, but you've also got to have the use cases, the understanding from the enterprise about what can be done with this technology. You mentioned the number of, you know, the millions of transactions per second to a more traditional enterprise. That might seem like, well, where is that going to fit with me? How am I going to leverage that? What are some of the things you're seeing in maybe some other industries, or how are you educating, you know, customers, potential customers, about the capabilities? Yeah, that's a great question. Because I do think now is the time when the traditional enterprise, the Fortune 50, the Fortune 500 is getting hip to this. But let's not forget, you know, Fortune, what? Where is Apple in the Fortune stack? Apple knows this. And Apple is beating the market, so is Amazon, by presenting personalized experiences. Anyone who's competing with Apple as part of their online store systems, or is competing with Walmart and Walmart Labs, they have great technology in all of these areas. If you're competing as a retailer with any of these companies, or with Amazon, you need these kinds of deals. You need real-time pricing. You need real-time stocking. Rolling those out requires an in-memory database today. So let's take the retail example. I'm a small retailer. Am I, should I go directly to somebody like Aerospike or to a similar, you know, vendor, or are there service companies that are kind of playing the middleman role? How do I actually go about and do that? You'll go to a service provider. So what we at Aerospike do is we get that technology out to a variety of service providers who are creating the kind of recommendation engines that then a smaller retailer might use as a service. So you're selling into those service providers who play the role of, hey, we're going to build a platform using Aerospike technology that does some of the things that Google's platform can do that Facebook can do, and then we're going to make that available to all their competitors so they can actually compete in this market. Absolutely. Cool. So Brian, how has the decision to go open source change the way you guys operate, change your business model specifically? How are you interacting with the developers, you know, saying differently in a more enhanced way? So open source is open doors for us across the industry. When we had gone into even, we found there was a big sea change, a big shift, market shift in talking to say the Fortune 100 and even Fortune 200 about any kind of platform. So we said, you know, look, you guys know you need an in-memory data platform, right? And they would go, yeah, yeah, we need that stuff. Then they would say, but it needs to be open source. They would just start about that. They would say, well, we're not willing to get trapped into any kind of data platform. If you want to be our data platform of the future, we need to be able to see the code and read the code. We'll still pay you money. It wasn't about money. And that was early when we were talking about cloud. It's not about price, right? It's really about trust. It's about understanding the technology. It's about seeing how that technology works. That's why these guys want open source. They're still happy to pay for it. They want a company behind it to support it. And I think a lot of the other NoSQL companies are seeing that too. It's been the dominant paradigm for new databases and correctly so. Right, okay, so the business model really hasn't changed dramatically. No, it hasn't. They're still paying you, but they just get open access to the code. And how about the contribution side? Is that picked up? Or what are your objectives there? One last quip was that open source is the new escrow. So in the old days, the old days, which is to say a few years ago, we would write all of our contracts that said, hey, you're going to get, in case of some unfortunate thing, say for example, Oracle buying Aerospike as a company, then that would trigger an escrow agreement. Now we just say, look, you don't have to worry about all that. We're open source. You're covered in case of any eventuality. That's really what they want is the new escrow. Second of all, in terms of contributions, we're not too worried about that. We have great software engineers. They understand our system. We get a few contributions, especially in client side, client side libraries. We haven't seen very money for the server yet, and that's okay with us. Really, it's about escrow. It's about trust. Open source is the new escrow. I'm Brian Bukowski. Thanks very much for coming on theCUBE. We're always really a sharp segment. Really appreciate your insights. Great, thanks a lot. Great to be here. All right, keep it right there, everybody. We'll be back with our next guest right after this. This is theCUBE. Live at Oracle Open World 2014 in Moscone. We'll be right back.