 Okay, we're back here live in New York City for Silicon Angles, continuous coverage, exclusive coverage of Strata plus Hadoop World. This is theCUBE, our flagship program. We go out to the events and extract a signal from the noise. This is the ESPN of tech, as we like to say, big sports fans, we love talking tech. This is Strata, Ground Zero in New York City for big data. And this is where all the action is for big data week and the industry is abuzz because it's the intersection of data geeks, coders, and business people, all intersecting into one melting pot of innovation disruption. Pleased to have on the program right now, guest Ben Werther, CEO and founder of a company called Platfora that just launched this week. Stealth startup, I got a chance to meet Ben earlier in the year. Got a little taste and he wouldn't show me the tech. He kept it on a complete QT on the raps and got a briefing last week and I saw you guys do the announcement. Ben, welcome to theCUBE. Thank you. Thank you, John. Good to be here. My co-host Dave Vellante is joining me as well and so I want to jump right in. I know you're super busy. I saw you getting hounded by the press. The rumor in the hallway is that you guys are the hottest startup here. I know you didn't win the startup showcase award. I don't think you were even entered. That's why I asked. Although Hedappt won, which is another good startup. We like those guys. They did a great job but you guys are being talked about as the hottest startup here at Strada. Mainly because your marketing is so good. It's not that good. The meaning is you're getting a lot of word of mouth recommendations. It's like really kind of low key but you're letting the product do the talking. So one, I want to drill specifically into that first but first set up for the folks what your company is, what you're doing, why you're just so disruptive and why the buzz is so high on platform. Absolutely. So we've been living in this big data world dealing with all this data. Hadoop is obviously a hot topic here. If you look back over the last couple of years, Hadoop adoption is exploding but the challenge with it is all these vendors are helping you set up a Hadoop cluster, get it up and running, do all the plumbing, all this talk about the infrastructure but when it comes to actually using, if you want to let everyday users do useful things with all that data, where do you begin? I mean it's really, really hard and this has been the missing piece. Companies have these pilots and they can store all this data that they were otherwise throwing away and if only there was a way to go do useful things with it and there's been a rash of vendors that have been announcing kind of connectors for Hadoop but for the most part that is something that doesn't really work. I mean the challenge is you're sort of, it's a whole different metaphor, it's a whole different architecture and so just poking at it using a legacy product hasn't been at all successful. So we're the first company that's really taken this view of how do you go from the ground up and really build a new stack that's natively about business intelligence and exploratory BI and analysis for Hadoop for big data in a way that lets everyday business users visualize, explore and most importantly get out of the business of having to build data warehouses and data mats and spend six, 12 month cycle times. So many stories of these early customers we've been dealing with where until now every new question meant a six, 12 month IT process. We want to get that out of the loop entirely. So one of the things we heard on theCUBE in fact the guest that was just on before you is the CTO of PlaysiteQ and using a technological cognitio which is an in memory thing but he basically just said Hadoop solve this big batch problem. But BI you need to slice and dice the data and that requires some schema and some benefits of relational database management systems and we were at IBM's IOD earlier in the week and relational database management systems are actually growing. They may not be great for no sequel but doesn't mean they're going away. So you seem to be doing this. So what is the big deal about that? How hard is it? What problem are you solving? Is it the you're doing unstructured data with Hadoop for the BI? Because from what I understand it's a schema challenge there. Yeah I mean it's a really good question and I think that what we've seen as an industry is you know doing it the old way doing it the relational database way it's like building an organized filing cabinet. I've got to do this six, 12, 18 months in advance organize the data but once it's in there it's very efficient to access it. It's just hey if I need to change the question I need to ask that's tough I've got to go spend six months iterating. Now that's- Hold on I'll say that again. So classic thing we've seen as for example you'll build a data warehouse with some data you'll organize the data in a very careful way and you'll roll it out to your users and the users will start using it and they'll say well but hang on I need to know more about, I see a pattern here there's an interesting point where something broke or some interesting trend changed. An insight they want to ask another question. They want to ask the next question down there what we call the next exploratory step. And that next exploratory step. The next big question is Cloudera would say. Exactly, exactly. And the challenge is doing that often means redesigning your database. I mean often it's the data you need to answer that isn't even in that system. We'll get back to you on that. And that's what that pain point is what in terms of time frame. That's six plus months. Six months optimistically, 12 to 18 months in a lot of cases really to get it solved. We had a sound bite on the cube earlier yesterday that said imagine using Google once a day and then you could only use one query a day or once a month, what would you ask Google? And they're like wait a minute I ask Google like a lot of things every day. So is that the same kind of thing you're talking about? Rapid kind of accelerating through Q and A kind of? Yeah, I mean drill this home really to make it clear. So I am one of the, there's a retail customer company we talked to right at the beginning of this and we said well hey let's say a user says hey this is the thing about customers that I really want to know but it wasn't brought through this into my data into my BI product all the way through the stack. How long does that take? And they started mapping out the series of steps. You think okay what could it be a week, two weeks, six to nine months, three different outsource firms working together, monthly risk mitigation meetings, 10 to 15 engineers, all these phases of work to get it through to be in the BI product. So there's no question that that's why Hadoop is exciting people but you're right you need to bring back schema and structure and you're seeing a lot of talk about that those kind of trends here at the conference. How about your product? Now I'll just set your website here it says making Hadoop amazing. Describe your product for the folks. So what we do that's different is it's not enough just to point at Hadoop and hope it's going to be fast enough. You know Hadoop is getting faster but it's getting faster in the order of maybe hours go to tens of minutes. You know it's not, if you have large volumes of data it's not interactive speeds. We're all about giving users sub-second interactive exploratory BI or completely web-based BI. And so there's this impedance mismatch between Hadoop and the speed it runs and end users. And normally you'd fill that with a data warehouse but that is that inflexibility point we want to get rid of that. No more data warehouse, the end of the data warehouse. So what we have is a complete stack that goes from any existing Hadoop distribution all the way up to the end user. Three layers. At the front it's a completely web-based exploratory BI product. Beautiful, fast, like super fast. Hundreds of thousands of data points selecting, interacting, exploring. You know, works on the web, working on tablet and phones and so on. That's, it's different, it's cool but that's not even the most interesting part. The real interesting part is that we create automatically aggregates and drive distilled Hadoop data into this scale out in memory layer that spans one, ten, a hundred machines depending on what you need. That, so it's automatically driving Hadoop. You're not writing MapReduce jobs. You're not having to hire programmers and developers. You're literally letting end users say, hey, I want to work with this, you know, video views data and I want to look at customers and devices and it's going to drive Hadoop to automatically build that into this in-memory structure that's super fast, super attractive. And then they just start building visualizations from that. Because it's this two layer approach where we're able to take the batch stuff and do it in kind of, you know, it's like the tracks lay in front of you and then you travel where you want to go rather than having to wait on each question, each step like you do today. So you use terms, data refinery, which is cool because we actually coined the term data factories on theCUBE three years ago with Abhi Metta, but it's a factory, right? So you got some data. I like this in-memory lens. So talk about this fractal cache thing. Is that, what specifically is fractal cache and this lensing concept? Absolutely. So a lens is like a, it's like a data mart, a dimensional data mart. It's something that is, but it's not, it's not something that I ever had to go manually build. It's automatically being generated out of the Hadoop data into this in-memory structure. And we talk about what we call fractal cache technology because that in-memory layer is also a, it's a complete distributed query engine. So it's driving the backend of this visualization environment. But the key thing is, so here's the perfect use case. I see something interesting in a visualization. I've got this set of, maybe a scatter plot and there's this weird anomaly and I want to explore that deeper. And I go literally select that in the UI. Today in most BI products, you know, they may say exploratory BI, but I click to drill down and it says, sorry, that's as deep as it goes because that's what we built into your database behind the scenes. In our model, without even leaving the visualization, you click a button. It says, I want to refine this further. And I say, here's the additional information I want. And I say, go. And it says, okay, working. And half an hour later, you know, it's running up behind the scenes. What happens with the schema? So what happens is actually, so the fractal cache is this notion of the schema evolves to make room for that new data. Essentially, it's flexibly driven. Essentially, it's a reflection of the underlying and loop data. Okay, so the people are going to not believe it. So I got to challenge you on this because, so is this working today? Absolutely. And you have customers? 10 beta customers and then we have a line of 15 to 100 customers that we're just getting ready to basically line up as soon as we have capacity. So there's a lot of interest. So I got to ask you, is the traditional data warehouse a dinosaur? Is it dead? Well, my keynote and my blog post were the end of the data warehouse. And so I think, I mean, look, yes, in a sense, I think that there's always going to be a role for the data warehouse. The way we see this playing out in our early beta customers is the traditional, the canned reports, the things that don't change, you need to do the same thing every day. That's the traditional data warehouse. It's traditional BI products that be, you know, business objects, micro strategy and the rest. But when you get to the world of exploration questions where you usually want to come in and say, hey, I want to follow hypothesis here and let me see what I can discover and then share out to the organization and discuss these insights and refine them. That whole world, the data warehouse is the problem, it's the impediment. And if we get it out of the way, we can build this much more agile. So you're building BI from the ground up with Hadoop? I mean, we're replacing ETL data warehousing and BI into the stack. It just sits on top of Hadoop and drives a seamless experience. Yeah, I mean, as you all know, I mean, the experience with the legacy data warehouses is awful. I've said many times, it's like a snake swallowing a basketball. People are chasing chips, they got one of everything. It's a patchwork. It's slow as hell. It's not really real time. Well, right, but the elapsed time is untenable. So what do you tell CIOs that are obviously struggling with this problem? I mean, should they be, I mean, you must tell them they have to be more aggressive about investing in the new stuff, but how do they go from where they are? So it's actually really, it's remarkably straightforward in a way, which is, I think part of the power of this and why we have 50 to 100 customers, we're basically waiting to use the product, is, so it's a two-step process. Step one, do you care about Hadoop? And so if a company isn't using Hadoop yet, you know, we give them time, but if they understand that the idea that they're throwing away all their data, that they're not able, that they're living in yesterday's questions isn't enough, then they want to start using Hadoop. They start investing in Hadoop. They start putting some data in there. Step two, they then try out all the existing solutions. They decide that they realize painfully that nothing works. And that's when literally we've had people, even with our stealth website, have been tracking us down, trying to get in the beta program, because there's no other way to do this today. There are a lot of vendors with a lot of talk, but fundamentally, you ask customers and none of it works and we're able to give them an experience where literally we're going in with our software on a USB stick, installing it within two or three hours. Those end users are doing subsequent interactive visualization, exploration. One of our early customers, big customer, I can't name them yet, but a really large customer, there's a woman in there who'd been using Hadoop for just a few weeks. She was a business analyst. She was completely, completely, you know, I mean, she's a really smart person, but didn't really know where to proceed, what to do, don't really get any value out of Hadoop. We installed our software, we spent a few hours getting her going. Come back three weeks later, she's now the expert on the team, she's training everybody, she's super productive. So the ease of use is a feature of yours. Oh, absolutely. And the rapid, rapid, rapid time to value, like literally a day. So two things to stick out, one, the time to query. So how fast can people recycle through this lens thing? So what you're saying is that you could essentially get new questions answered without massive schema changes. Yeah, I mean, most, I think the vast majority of things people want to ask, there's a lens already there, so for them it's just, it's sub-sick and they literally, boom, it's there. It's like that fast. If they want to drill in, in those cases where they really need to go further, then they want to have a lens refined built for them, that's just a point and click operation. It may take, depending on the Hadoop cluster, five minutes, half an hour, maybe an hour if there's a lot of data. But that's as long as it would have taken to query Hadoop in the first place. Instead of just building one result, they've built out this entire interactive data map that's going to support the next day or two of work that they're gonna do. So your marketing, I was just looking online here, it says, Hadoop with no ETL or data warehouse required. Absolutely. That's the bumper sticker. Yes, yes, yes, yes. So that's pretty scary. People like, I mean, you get any hate mail from data warehouse vendors, A&A, your bullshit, you know, come on. Well, I get those, my favorite tweet by another, a big vendor that I won't name who's actually a major sponsor here. Some folks were talking about how- That would be Informatica folks. No, it was a different one about it. That's another one. Who was talking about, they thought this was something from The Onion when they read my blog post. It's too good to be true. This is this copyright, you know. This idea of the world is changing. I think we've seen a lot of people come by, you know, honestly taking photos of our booth, of the UI because they understand something different is going on here, which is really cool. Yeah, I was talking to one of your employees at your reception the other night and he left a really cushy job at a big company because he said this is magical story. You get a lot of good buzz, congratulations. Can't wait to see the product and get a demo, come by. So at our booth, you can come by and see it. Yeah, I want to talk to the customers. I mean, to me, my big thing is, we were talking earlier, customer validation ultimately trumps everything. Absolutely, so I want to come back to that a little bit. Ben, you made a statement about, if the guys aren't using Hadoop, that's not really our target. If they're happy living with yesterday's questions, that's cool. I mean, what percent of organizations that are living with yesterday's questions are going to be out of business if they don't make a transition, yeah. I think that's the magical thing here, which is you think about, it's not about, I mean, cost of ownership is one thing and we can argue all day long why this is a much, much more cost effective and scalable way of doing things. But the thing that blows people away is, there are business analysts, we literally say to them, well, what questions can you ask? And they can't ask anything that wasn't baked in by an IT person three years before. And they're living, they're running their business off a set of questions that are obsolete. And so the idea that they can literally sit down this afternoon and begin to answer questions and ask questions about the business in a new way, that's, I mean, I don't think anybody's even been on a measure or begin to think about the implications of that for the business. We have been on this meme for a while now, just basically the last seven, eight, nine, 10 years have been about, like you say, TCO, cutting costs, doing more with less, cutting IT budgets and so forth, but we see the productivity impacts of big data and other technologies is really being enormous and many people in our community have forecast, look, IT spending is actually gonna, it's gonna bottom and then as a percent of revenues and if you're not spending more, you're gonna be left behind. Absolutely. In the right places, obviously. We did a great case study to your customer question, a great case study presentation jointly with Capital One Labs right here just a few, couple of hours ago and so we had a packed house. People, it really interested in, you know, how are we working with Capital One Labs as one example doing some very interesting analysis of, in this case it was about smartphone usage in terms of credit card purchase behavior and they were understanding their app adoption and they were able to very quickly reach some insights that were, you know, this is a very, very first. We showed a Microsoft demo earlier and my tweet was, you know, Hadoop needs to go and get ready for the mere mortals and normal people, like not, get some beauty science majors. Absolutely. Where everyone can ask the question. So with that, what's the next milestones for you guys? Obviously, you launched the company, you got to roll out and get the product out there and not do memory sticks but go out and close some big business. Given the progress, you're gonna accelerate your staff, accelerate your sales. What's your strategy? Just hold on and hope you don't fall off the rocket. I mean, or. I mean, really themes. I mean, one is we're in beta now, the product's looking amazing but we've got to get to GA so that's something we're targeting. I mean, we're on track for Q1 for that. We're building out, we have relationships with all the major Hadoop distribution guys, a lot of the SIs. Everybody's very, very excited but we're just trying to, you know, figure out, you know, how we engage so that we can maximize and accelerate because they're really excited. Their sales forces are actually trying to call us in and say, can we use platform to help our sale Hadoop? I mean, it's a much better story if you could show why it's useful to people. So there's that, I think, you know, we're obviously, we're hiring, we're hiring on the technology side, you know, engineers, sales engineers, sales people across the board. You got Andresha Horowitz as your big VC, right? So Mark Andresha's firm and Ben Horowitz. Absolutely. Anyone else in there? So Sutter Hill Ventures, also Incutel, which is an investment arm of the Intelligence Agency. They're gonna ask better questions too, we'll get out and look at it. And then some other, a lot of other, there's Angel Angels, Data Collective as well, and that team. So, you know, the Andresha Horowitz group have been fantastic. How do you feel right now? Do you feel good? I mean, it's been a fantastic show. I mean, just seeing the reaction, seeing the sense that people get, I mean, we knew we were doing something different, we knew that it was gonna be a good reaction, but I think that I'm veiling it. Just seeing the buzz, I mean, seeing that, you know, I mean, I had a 500 keynote right after Mike Olson. Mike did a great job, but we were being talked about as, you know, if not equal with, you know, Clodera, like the, in that, you know, either number one or two in terms of the most important stories of this event. So this has been a really great show. Yeah, I think so too, the buzz is fantastic. I mean, disruptive and innovative, but real, practical, and you really welcome the talk there. Absolutely, people at our booth playing with the product, seeing it in action, they're testing it right there, so it's not, you know, it's a real product. Causing a traffic jam in the 3,000 people in that small little hallway. Absolutely. Okay, Benworthy, congratulations. You're the hot start up here at Strada, Hadoop World, congratulations. Quick final question, Hadoop Distro, any preference of you with Clodera, Hortonworks, Republican, Democrat, Green Party? The great thing about what we do is, you know, we make everybody, we help accelerate the entire ecosystem. So we work with Clodera, Hortonworks, MAPR, Amazon EMR, we're talking to the Green Plum guys about, you know, how we do more with them. You know, we have a lot of friends that are nice and Switzerland, you're Switzerland in the big data world. Absolutely. All right. Okay, Ben, congratulations, start up. I know it's a lot of hard work to launch a company, congratulations, great success, and you deserve it looking forward to seeing those case studies. We'll be right back. This is platform hot start up here inside the cube. Always bringing the start up action, that's where the signal is right here, so look at the angle and wiki bomb, we'll be right back with our next guest.