 Live from the San Jose Convention Center, extracting the signal from the noise. It's theCUBE, covering Hadoop Summit 2015. Brought to you by headline sponsor, Hortonworks. And by EMC, Pivotal, IBM, Pentaho, Teradata, Syncsort. And by Atunituandisco, now your host, John Furrier. Okay, welcome back everyone. We are live in Silicon Valley in San Jose for Hadoop Summit 2015. This is theCUBE, Silicon Angles Flagship Program. We go out to the events and extract the signal from the noise. I'm John Furrier, the founder of Silicon Ames. I'm joined by guest Peter Goldmacher, who's the VP of Strategy and Market Development, Aerospike, formerly theCUBE on the panel as an analyst at the Big Data NYC, which was a smashing success. You had great commentary. That was the most engaging panel. Lights were a little dim, but the content was smashing. Welcome to theCUBE. Thanks John, great to be here. Big Data, you've been there, you've seen the different perspectives of this as an analyst. Now you work at Aerospike as a strategy. What's the chessboard look like? I mean, strategy is about the chessboard. Three dimensional chess, you're looking at it, you got this, that, going on. What's going on in this industry of Big Data? Obviously Hadoop is open source based. Open sources has been validated and continues to be legitimized every day. Tons of innovation, softwares kick ass. Big vendors are playing in here. Apache, super successful. What's going on in open source and in the Big Data? Yeah, it's a great description. It's not just a chessboard, it's a 3D chessboard. So when you look at a lot of the technologies that exist in the market, a lot of them are lower cost replacement alternatives for what has historically been proprietary software dominated by a lot of large vendors with enormous pricing power. So a lot of what you're seeing in the open source community is a reaction to abuse at the hands of the established order for the last 20 years. So a lot of Hadoop opportunities and NoSQL opportunities come about because customers say, hey, there's a better, cheaper way to do things. I've been restricted to what the proprietary guys sell me, now I have new opportunities. But the incremental layer, and what I find so exciting is these technologies are actually enabling new, new things. So if I think about compute, we've got a new way to do old things better. But the technology, both software and hardware has come so far that they're now new, new things you can do. And a big- That are disruptive to the guys with the pricing power and market power that could have wielded that power even 10 years ago. They're disruptive to everything. Now that they're becoming powerless in a way because open source is so powerful that they have to join. So open source is, open source is a separate topic. I think about just the technology. Let's not worry about who's writing the code or who's making a commercial available. Let's just talk about the technology that exists today and the things you can do with it. That's an amazing story. Open source is a whole nother wrinkle that- Let's go down and say, let's talk about the new, new things. So expand on this new, new things. You're talking about like internet of things. You're talking about new and analytics. So I'll use concrete terms. When you look at the marketing, go-to-market strategy for companies like MongoDB and data stacks and couch, their primary message is, hey, we're an oracle alternative. There are things that you built an oracle or any relational database. And the reason you made that choice was because you had an oracle site license and it was the best technology available at the time. But now there's new technology. Technologies move forward. The price points are very attractive. They're often free. And you actually have an opportunity to build, use technology to build applications and that underlying technology is more appropriate for that application. So in Mongo's world, they do a lot of content management apps. They do a lot of product catalogs, a lot of web-based apps. And this is better technology for these use cases. Oracle is typically overkill for things like this. And data stacks and couch have much of the same opportunity. So their primary go-to-market is, we are an oracle alternative. One of the things I like most about being at Aerospike, which is, has raised a fraction of the funding that Mongo has and is in a smaller, but I think faster-growing market is, we can help you do old things better and I can give you some concrete examples. But if you look at the preponderance of our use cases, these are in brand new categories. These are in categories like ad tech. Ad tech did not exist 10 years ago. Ad tech exists because now you can take advantage of enormous gains in software and hardware and have a make immediate decisions on a mix of historical data that you can capture in Hadoop and immediate data that comes into the app from whatever you're doing on your cell phone. So when you go to a website and you go to the Nike website and you go to the Puma website and you go to the Adidas website, these ad tech guys are saying, hey, what's John up to today? We know who John is because we built a profile of John in Hadoop and we know John is a 20-something guy and he's got a wife and he's got two kids and he lives in this area so we've got a profile of who John is. But if he's going to the Nike website and the Adidas website and the Puma website right now, if we serve him a banner ad for a big five sporting goods or a Models or a Dix or whatever this, the probability that he clicks on that link is high and I can sell that banner ad for a much better price. Re-targeting. Re-targeting, but it's re-targeting with much better visibility than you've ever been able to have. So you're basically identifying intent in real time. So our customers are and the way our customers are able to identify intent is because they've been able to collect enough data and make an immediate decision based on all that data. So that's not an old thing. That's a brand new thing. It's funny, the curse of our success in ad tech is it's still such a new industry. We have these customers with amazing workloads. We have customers doing two million transactions a second on 150 terabytes of data. And if I gave you their name, you would probably never have heard of them. So we've got these industrial case use cases that don't have the brand recognition of some of our later clients that saw the use case and then said, okay, we get it. So in that case, you guys are providing such high performance that you're giving them an edge in arbitrage of the new capabilities that impact revenue. It's kind of like a financial trader who wants to have that second edge advantage over the other guy. I mean, Walmart, for example, I know, looks at this stuff heavily, like literally the performance of like in seconds change millions of dollars in revenue. If you think of the business as a commodity, the way you get an advantage in a commodity is speed. So we offer our customers speed and speed at scale so that the bigger your opportunity gets, the bigger your data set gets, you don't have to, it doesn't slow down your opportunity to take advantage of it. That's a great point. I'm going to have to tweet that. Commodity speed is, so let's talk about speed. Let's talk about open source for a second. Let's get down to that. What's going on open source? What's your thoughts on open source? Yeah, so my thoughts on open source trace back to my time as a research analyst on Wall Street. And what I find so interesting about open source is it's not clear to me that for anyone other than Red Hat, it's a commercially viable model for other than a very small subset of the companies that are currently open source. I think the genesis of open source comes from the reality and the reaction to abuse at the hands of the proprietary guys, right? Oracle has had so much pricing power in its install base for so long, when you have someone come along and say, hey look, this stuff is free as long as you, it's free and it's community built and you can have it for free, the pendulum swinging the exact opposite way of pricing power from the vendor is no pricing power from the vendor. So one of the nice things about open source is you get pretty dramatic adoption because if it's free, it's someone certainly willing to try it because you don't have the overhead associated with a big price tag. What do you think about the different approaches of Hadoop? Obviously there's a post Hadoop world. You got Hortonworks at Cloudera. Enterprises seem to be attracted to on a large scale, much more of a Hortonworks, but yet they see the Apple deal with Cloudera, rumored to be north of $50 million from licensed deal. So the challenge with open source now is, I don't know how many open source companies have been funded by the venture community. It's got to be well over 150 and the amount of money they've raised is enormous. So what you have is these open source projects that aren't really open source in the sense that the community is contributing because the majority of open source companies have a commercial entity behind it that write 99.99% of the code. And that's true from everyone from Mongo, certainly Aerospike, that's the dynamic we find ourselves in. So the challenge is, how do you make these commercially viable entities? Look across the open source landscape. I don't think anybody makes money. The only one we can really talk about is Hortonworks because they're public and they have to follow their financials. They don't make money, they're not going to make money. It's going to take a long time. So as long as... How do they make money? I mean, the railroad analogy, Herb used electricity was the word that Rob Bearden used. Well, I think... That's commodity one and it's utility. It's a monopoly. I think the way they make money I think the way they make money is if you become the standard and you become bigger than everyone else and you become the vendor everyone goes to then everyone else just kind of falls off the back of the bus. So what Hortonworks and Carrera are doing is they're saying, we want to be the one. So they collect enormous sums of money either from public markets or private markets and they build out these massive distribution markets on a complete field of dreams. Thought process that says if I build it, they will come. If I can establish myself as the leader, however I choose to monetize I'll be successful because I've got heft in the market. So what you're seeing is these guys if you look at their financials their spend on sales and marketing relative to their spend on R&D is massively skewed towards marketing and not towards product. And we'll see what happens, right? I mean, there's only a few field of dreams scenarios. I mean the .com bubble was built on field of dreams, right? That's a big deal. How did that work out? It didn't work out very well. Yeah, but I mean that's a all-in go big or go home kind of way. I mean. It is. And simple math would tell you most of these vendors are going to go home. Okay, so of Hortonworks and Clouder which one do you like better for the I'm going all in and I'm not going to go home? Well. Or can they both be the guys? They have pretty different strategies. I think they're both committed to keeping the open source core vibrant and Hortonworks model is primarily a support model and Clouder's model is proprietary software that you sell software on top of that open source core so they're augmenting the core. I don't know. I don't know. They're both interesting enough and it's tomato tomato, right? Some people are happy to pay for support and innovate themselves or layer in technology they choose. Others want a bigger part of the solution and they'll buy it all from Cloudera. What we find at Aerospike is there is a very high correlation in our customer base between someone who's using Hadoop and we're agnostic whether or not they pay for it or they're using open source and they're likelihood that they use Aerospike. Almost all of our customers are actually Hadoop users so we care a lot about the Hadoop ecosystem remaining vibrant and the situation is what you can do in Hadoop is you can, the value proper Hadoop is fantastic. I have a very low cost way to take in a ton of data and find the value in that data but once you find the value in that data what do you do with it? We give every Hadoop user the opportunity to execute on that data immediately so you're not constantly building profiles and then taking three days, six months to get the value from that profile. We're saying give me a profile for John. Give me a weighted. You're instant pipelining the data to the transactional. Well our customers are writing applications that take in Hadoop data. You're enabling your customers. Well we're just a database. So I'm going to ask you about Spark, okay? Brian, the founder of Aerospike, when I first started talking about this in-memory thing I was like, man that is badass. Almost too early, right? It's like classic innovator, right? Way too early but now it's mainstream. Spark's on the, people are buzzing about Spark but that enables, again, Spark to me highlights the Aerospike Viper, am I right to kind of connect those two in concept or are they different? Definitely connected in concept. So in-memory is another kind of big word now that's very trendy but there are different flavors in-memory so let's be specific. So what Spark does is basically because it runs in-memory you can put more compute behind it. You can do what you normally do in Hadoop faster. So if you want to run a batch, let's say in a traditional Hadoop environment it takes 10 hours in Spark, let's say it takes two hours. Which is great because it means you can build your profiles faster. So you've got actionable data that you can build that profile faster. But when you talk about in-memory you've got to be specific or you're talking about RAM or you're talking about SSDs because there's an enormous price performance difference. So when you- RAM is what everyone, that's the killer. Well RAM is the killer but you pay through the nose, right? So when you talk about SAP's HANA and you talk about Exadata and you talk about a lot of these guys that say they run in-memory what they're saying is we run in RAM so we take our code that's always existed we put it in RAM and we can run that code 20 to 30% faster. It's a distinction between RAM and persistent memory which is SSD in our flash. Yes, so what they say is put more hardware power behind your traditional software code and we'll make it run faster. What we do is we say no we've written a completely different kind of database to run explicitly in SSD. So you're not doing all the normal database operations in aerospace you're running right on SSD so we don't talk about 20 to 30% performance improvement we talk about 100 to 150 times performance improvement. I got to ask you a personal question about Aerospike. I mean I have a personal connection with Brian I love that team over there because I like them because they're great engineers and they got a great value proposition but they're also underfunded relative to the big hyped up markets out there so kind of root and form to kind of kick ass. You have a lot of choices to go why Aerospike? I mean what made you join the company from a TAM perspective? What dots did you connect besides being enamored with the technology? What was the next level? So in my experience as a research analyst on Wall Street I met with I don't know how many hundreds of companies and you can generally divide a company into one of two categories. Shitty product, great marketing or great product, shitty marketing. And Aerospike for me was if you look at their install base and their customers and the things they can do it is a great product. And the opportunity to join John Dillon who's the ex CEO of Hyperion of Arbor of Salesforce.com with Engine Yard. They were just at Engine Yard. And work with John who is a go to market guy and have that experience and take what we know to be a great product and bring it to market. You just you can't say no and to compound that decision even more my commute is 51 miles each way every day and I'm in the office every day. So it's incredibly I mean, you're not mailing it in. You're driving 51 miles every day. 51 miles each way every day. That's a hundred. I'm not a math guy, but that's 102 miles. So the opportunity Check those suspension force. The IRS will be like, okay, let's set those mileage calculations. They better get you a car. So, you know what? I agree. Or a driver or a helicopter. Hey, Uber, you know. Yeah. So the opportunity, so the way I look at the opportunity is look we're not an Oracle alternative. We don't go to market as an Oracle alternative. We go to market for very forward looking developers that are thinking, look, I've got an opportunity if I can have unbelievable performance in my database what are the kinds of applications I can build? So we are in a way the commercial version of the technology that Google and LinkedIn and Facebook use to build these businesses. These are data businesses. These are the guys that said I have an idea but I can't execute on the idea unless I have unbelievable technology and they built that technology that's all proprietary. So you've got a great engine. You've got to get this Ferrari built out, shine it up, take it around town, top down kind of thing. So you got to kind of, you know, the engine's solid. Now you got to kind of put the bells and whistles around it. So the go to market for you guys is just run like the wind on the go to market sales and then more top level marketing. So one of the things, one of the decisions that was made when John and I made together and I think it's a decision he let me make, I think he was already there is. Look, if I have a dollar to invest, do I want to invest in a great product or do I want to invest it in telling people about a mediocre product? So one of the things I did when I got there was I took the marketing, so let me take a half step back. John Dylan says, hey, come join me at Aerospike. You can be my strategy guy. You'll work for me. You don't have to, you won't run anything. You'll be an individual contributor, which for me was love it. That's exactly what I want to do. And then about four weeks after he says, you know what? Why don't I put marketing under you? And I said, well, I don't really want to do that. And John said, I really wasn't asking. So when he put me in charge of marketing, I said, boy, we spend a lot of time trying to build awareness in a market that's not really our market. And we've raised a 10th of the funding that Mongo has raised. We've got the smallest megaphone in the echo chamber. Let's get out of that business. Let's fall back on our strengths which is product. The budget I had allocated, I want to give back to my product guys because there is no better marketing dollar than a happy customer that tells his friends, boy, this Aerospike stuff works. Yeah, the referral is the key. Because customers are talking to each other right now. Absolutely. So what we've done is we've, we have completely regeared our marketing to go after developers and development teams because our thought process is if we can give these guys a great experience, then they become our advocates. So we have a, what we think of as a virtue of cycle. And check this will value the product too because they like the Ferrari. It's a Ferrari model. You guys have a Ferrari go to market. You got to drive it around town. You got to have the red one out there, top down, doors swinging out. You know what I'm saying? That's kind of where you have the high end product. Well, we look at it, we look at it as an awareness to advocacy cycle where we drive awareness and then we eventually make our customers our advocates. And that's our best, that's our best marketing. Okay, we got a break. We're getting the hook here from Leonard. So I got to ask you one final question. Yeah. What's next for you guys? What's next for this ecosystem? Oh, I think for us, you're going to probably start hearing about some pretty industrial strength use cases and things where you say little company, enormous customer, what are these guys up to? We've got a couple customers in the pipeline we haven't publicly announced yet that when we do people will say, wait a minute, I should probably start paying attention. And in the ecosystem, I would say everybody is investing aggressively and go to market and they're spending a ton of money on this stuff and the burn rates for all these companies are extremely high. And as long as the venture community is active and vibrant, then I don't think much changes. If that's dynamic changes and capital gets a little bit harder to come by, I think we're going to see a couple of pets.coms and webvans in the open source private company space. Great, we've got some great chats, comments on the crowd chat. Appreciate the comments. Certainly getting it out there in the trenches is the way to go. We love that strategy. It's theCUBE, of course. We're out there sharing the data with you guys out there watching. We have Aerospike on theCUBE here at Hadoop Summit 2015. Move right back after this short break.