 Live from Boston, Massachusetts. Extracting the signal from the noise. It's theCUBE, covering HB Big Data Conference 2015. Brought to you by HP Software. Now, your host, John Furrier. Okay, welcome back everyone. We are live in Boston, Massachusetts for Silicon Angles theCUBE, our flagship program. We go out to the events and extract the soothing noise. I'm John Furrier, the founder of Silicon Angle. Getting down to day two, wall-to-wall coverage here at HP Big Data Event, special presentation with theCUBE. My next guest is Joseph Yen, Senior Manager of Business Development, Vertica, HP Software, and Colin Zima, Chief Analytics Officer, VP of Product at a hot startup called Looker. Kind of a growing startup, you're really not a pure startup anymore. Get over 300 people, welcome to theCUBE. Thanks for having us. Welcome to theCUBE, good to see you again. Good to be here. Okay, so guys, give us the scoop. All right, you won the hackathon. Looker had the big trophies out, parading it like the NASCAR driver. Gotta celebrate. Champagne popping, didn't spray the crowd. We're going to drink out of them later. What's going on? Give us the overview. What's going on, Looker? Tell us about the hackathon. Give us the overview. You want to start with the hackathon? Yeah, so on the hackathon, we got a data set of all the New York taxi cab rides in 2013. We pretty much used it as an opportunity to show off the tool. We just dumped it all in there. Found some pretty interesting stuff. Just some crazy insights that pop out of that. Like, if you're a cab driver, actually tips peak at 5 a.m. in the morning. That didn't make any sense. So we dug into it and it's actually, it's when all the airport rides happen. So if you're a taxi cab driver or you want to start driving for Uber, I guess get on the roads at 5 a.m. So that's why they wait in those long lines and that's why they're yelling at all the Uber drivers. Hey, you're taking my tips. Now you guys have Uber as a customer, right? Yes. So you have a little insight. So did you game the hackathon? I got to ask. No. Why'd you pick taxis? We got the taxi cab data from HP. So this was, you know, blind, no specific expertise here. Did you present the data? Did you rig it for them? Maybe. No, the product works pretty well on Marketplace data. It's a flexible tool that really lets you get at dirty data, like sequential rides and turn it into things that you can actually use. I was only kidding. But let's get the institution jumped in and let's get in the product. So go over the product. What's the key value proposition? What's the use case that you guys are crushing it in? Where are you winning? And why are you winning? So the company was built on a couple major theses. I think the first and kind of the place that we differentiate the most is that we're entirely in database. So where a lot of BI tools are actually sucking data out of the database, we saw databases getting much faster and more performant, all this good stuff, like the verticals. And we said, let's leave the data there and actually start just issuing queries directly into the database. So it's all in database. It's entirely cloud based. And in between, we're actually able to transform at query time. So you're not reshaping your data in the ETL pipeline or in the back end. It's just- Great for unknown queries coming out of the woodwork. That's exactly right. It's an exploration tool first and foremost. And we found that that's what a lot of companies want to start doing now. It's actually just getting at the raw data, dump it in there and just go attack it. So Nate Silver's keynote, fail before you succeed. That's essentially iteration, call it terribly, lean startup or agile. Same mindset, right? Or with data. That's exactly right. Don't spend all your time reshaping the data to answer questions that you might not know about yet. Just get it all in there and start answering those questions on the fly and just don't move the data around. Just get at it. So talk about the database piece. Are you in Vertica? Do you have your own database? How does that work? So we essentially connect to any database that can write SQL. So that goes from the row level, MySQL Postgres, up to the MPP databases, the Vertica's, the RedShifts, and even the SQL on Hadoop. We've started working with some of those companies. The analytics there is not quite as mature yet, but really anything that can write SQL, Looker can write SQL too. So, okay. So are you guys primarily a visualization piece or are you just a query engine? Do you, does it matter to you guys or to customers? How would you describe it? That's kind of the question. So people are trying to do all of those things. It would be naive to say that we could not be a visualization tool or a reporting tool that actually is doing the things that we're doing. It's kind of, that just goes with being a VI tool. You have to do that stuff. That's just the cost of entry. We focus and we differentiate on the explored experience. So at some level, it's hard to differentiate being a dashboarding tool. It's almost like you don't want to pick. Yeah. It's like the market kind of figured out, but you do become what you're known for eventually, right? Yeah. So that's kind of the trick. And that's exploration for us. We want explore to be what we're known for. And we want to do the other stuff to enable you to have one tool. You get more breath in terms of opportunity because if you become a one trick pony on analytics, that's exactly right. Who wants to be a one trick pony? Yeah. So how does this work with you guys? How involved they with your ecosystem? I mean, you guys, you know, I mean, I'm super impressed with what I'm seeing with startups. I mean, you got tier one VCs funding companies with Vertica OEM. That's pretty, I mean, you know, 10, I'm old and 10 years ago, that never would have happened. It would be like build your own, don't bank on any vendor, they'll screw you over. That's the old days. Now the new days pretty solid. No, absolutely. I think Vertica, you know, historically has been really successful as a platform play working with a wide ecosystem of different partners where we can offer a variety of solutions built from best of breed components. You know, we do, what I would admittedly say is a fairly well-defined set of things and we try to do those very, very well. And we work with partners like Looker who are architected to take advantage of a lot of the performance and the analytics capabilities within our database, but then be able to surface it in a way that works really well for, I think, how modern business analysts or data sciences users would use it. You know, you want to move the data closer to the business users. You want them to be able to explore. Let's just pull a mic stone breaker. People want the data now. They want it fast. They want it relatively easy. That's exactly right. VI tools, I mean, who wants another VI tool? I mean, that's like, we're hearing that consistently across this show. Data warehouses, that's Hadoop, that's Spark. I don't want another business intelligence app. I want stuff when I need it. I'm an analyst. I need data. I need to visualize it. That's kind of what we're here. Answers, people want answers. They don't want tools. They want answers. So tell us, how does that happen? Where are we on the spectrum of making that happen? You mean just in general? Just querying data. Hey, you know, kind of like data just happens. I mean, right now, databases are fast enough that you can just dump a bunch of data into a certain place and if you have an analyst that knows how to get at that data and answer questions, they can draw real conclusions quickly. Like, the hackathon's actually a great example of this. We got this Taxicab data set and two hours later, we could probably go start, help run the New York Taxicab system better than they're doing it now. And it's just the ability to take all the raw data down. You're not angel-less. You get that funded like that. There you go. No, but this is the point. This is the new opportunity. There's a ton of opportunities out there. You innovate on one piece of the value chain. If you kick ass so bad, I mean, so good at it, if you're a bad ass that's so good at it, I mean, think about that. One little innovation. It's completely true. Like, we like to think that we're more effective selling the tool by actually helping you answer questions that you need to answer, solving problems for you, than we are selling the tool. We'd rather just help you solve business problems. I'm sure if I was the looker management team, you're like, hey, we should start an adventure capital fund just to fund the people using Looker. I mean, that would be like an interesting, not that you do that. And it's literally they're doing well. That is a dynamic where you're enabling. That's a platform. When you're enabling innovation, that's a platform. And I think there's another important kind of thesis that we had building the company. A lot of tools now are trying to do predictive analytics. They're trying to solve all the business problems for you. I think at some level that's an unrealistic expectation that just a tool can magically make you money. We very much said, let's make the data analysts more powerful and let really smart people be 10 times as effective. And it seems to be that's what young companies, they agree with that thesis. And they're hiring smart people and just giving them the tools to go answer things really quickly. Yeah, and I think the myopic view that these companies take on the solution side is, and I'm seeing Twitter do it with how they're curating data. I mean, they're making decisions on behalf of users that they don't even know what their orientation is. Just because they think Justin Bieber's going to have something's hot and they can do it on the long tail at the top. But I mean, I have no idea from dorm room to born room what people are interested in. And so you have to have a system that is so adaptive, they have to be valuable for everybody, without any context. So I think that's where I see the companies making mistakes is that they try to do these canned prepackage apps. And it's like a general purpose clearing engine. Yeah, okay patterns, people don't want to see the weather. Okay, I get that. Or you know what I'm saying? Well, if it's that easy, can it really be that valuable? It's kind of what it comes down to. Like if you can just pull it out of a box and have it be perfect, then it almost means that it was so trivial that you should have been able to figure it out in the first place. Are we just getting to just a search engine? I mean, isn't the end game just Siri voice activated or text based query answer? Maybe. I mean, isn't that kind of like the Nirvana? I mean, what is the whole, where's the promised land and all this? I think for certain people, they're looking for things like that. It's going to be tough to do that correctly. Because if you ask for revenue, there's 10 types of revenue, which one are you asking for? That probably is more of a mess than it actually seems on the surface. But I think that is what a lot of people want. A good data analyst is going to basically throw something out there, really fumble his way along. He's got a clue as an outlier. He's not sure he wants to have a time commitment, but we'll poke at something. And if it yields some information, okay, I'm going to double down on that. So like we're constantly poking at things, right? But that's the workflow is that it's really, it's that analyst just being enabled by all the stuff. The thing that might have taken them a week, now they can do it in an hour or 15 minutes. Yeah, if they're indifferent. Hey, I'm going to, you know what? Screw it, I'm going to hit the happy hour. I'm going to go home early. There's too much work. I'll be here till 10 before I see any value. Versus, wow, I'm going to skip happy hour. I just, I'm on a track. You know, you get vectored in on that. Definitely. Yeah, that's what I'm seeing. Do you guys see that too? Definitely, definitely. Okay, so now take me through the dirt stream that, a data analyst and a data scientist. Stonebreaker kind of teased it out on his keynote, throwing out the haymaker. So I'd say the data analysts are answering questions that might take you an hour or a couple of hours, and data scientists are trying to answer questions that'll take you a couple days or a couple months. A little bit more programmatic, a little more deeper penetration. Definitely, like I think of a data analyst as pulling out insights and the data scientist turning those insights into things that actually get implemented in the business pipeline. So like data analysts are still kind of working at the top of the pipe, where they're just attacking all the data. Data scientists now are pretty close to engineers. They're almost writing production code. They're building models that turn into the search engine. Things like that. I think data scientists, I'm old, so Fortran was the language you used for instrumentation back in the day. It's specialized, no one really likes it, but it was effective, and then the language just got better. The early days now seem to be that way. Yeah, Python, I mean, it's easy compared to what Fortran was, but the point is, it seems to be getting easier. Oh, for sure. Do you see that as well? Definitely. The tooling is just so far beyond even what it was five years ago. Just from the easier stuff, like querying, the things that Looker's doing, the exploring. But now there's whole stacks just devoted to making data scientists more effective. So they're saving things. They have workflow management tools, and then putting that code into production is just readily available. Okay, so let's talk about the HP relationship. So what's your relationship with HP, Vertica? I mean, it's so funny. Stuff comes out of the woodwork over the years. That's my third year doing this show. Oh yeah, we had Facebook, we had Zing, we had Twitter, we have Apple. I mean, maybe I'm spilling the beans a little bit here, but Vertica really nailed that high performance. The column of store really seemed to be the home run. Is that what you guys are using? What's the relationship? What products are you guys using? So we work with all sorts of providers. We find that people that use Vertica are approaching problems in a way that requires something like Looker. If you're making a decision to buy Vertica, it means that you're appreciating the analytical functions and things like that. You have a lot of data. Yeah, but when it comes down to it, your BI tool is sucking all of that data out of Vertica into their own backend, and you're losing all of that tooling that you're paying for, where we're sitting directly on top of the database. So if you want to use all of that analytical functionality, we're meshed a little bit better with the purchaser that wants. All right, so I've got to ask you a question, Colin. This is good. I'm getting word for Joe in a second. I think it's you in a second, but so here's a fear that I hear all the time. Man, I got my data. I did some H-Base over here. I got Hadoop, and I've got DynamoDB, I got stuff on Amazon, Tickets and Asks. We're somewhat funded. Why do I want to just dump my data into someone else's platform? That's a legitimate question people ask. So when is the point to do that, or is it a staging area for the unstructured, or what? We don't ever think it's the time to do that. People building companies, or big companies. Or I got an investment in X. So what's the answer to that? How do you stop that fear? What's the benefit of someone saying, hey, look, you're not really losing control? Right, I mean, it comes down to how open or closed the platform that you're putting the data into is. Like, some providers are just more comfortable allowing you to move in and out, and there's tools being built to enable things like that. So there's tag managers that sit behind all of your event streams, and they'll actually consolidate all that data so that if you want to go from provider A to provider B, that's really easy. It's kind of thinking about that proactively, rather than it being a problem that, the second that you need to switch out of a certain provider. What's your take on that question? No, I think- A little biased, but that's okay. We'll accept that answer. It's certainly an issue, something we think about it greatly. I mean, we don't want to have to tell our customers, oh, you have to move everything into Vertica, right? You have your data in a variety of different places, and Vertica, to different degrees, beyond us, we'll be able to work with it. We could work with data within Hadoop, within certain file formats within Hadoop. We could also, we're also working on a variety of different ways to interact with a number of other types of data stores. Well, you guys are flexible. I mean, I put you in the spot there, and you get the company answer, but I think I would say is that Vertica has choice. There's no real lock-in. You don't like it, you swap out. There's certain architectural and performance compromises or things that may work, may not work in certain scenarios, and depends on the application or the use case eventually, right? If you need some two-second response, then you may want to think of materializing into Vertica, but for exploratory work, we're being able to scan a wider range of data and trying to understand it, and maybe just trying to find the nuggets, then we could certainly work with a large number of data sources. I mean, it's a question we hear all the time, and I think the answer that kind of comes back, and I'm kind of paraphrasing it now, just to kind of put it into words, is it kind of depends. I mean, where's the value? Whoever can provide me the value, right? So if I'm hoarding my data out of fear of someone else is going to steal my value, there's the values in the data, right? The data is in the value. But hey, if Joe and Mike have better products on visualization, I don't, I'll still hold my data, but I'll put in whatever system gets me value. I think that's kind of where I see it going. You guys, how do you see that? I mean, do you agree? I would tend to agree. I mean, when it comes down to it, switching between a lot of different tools is not that big of a deal. So you really can actually just find the thing that works best for you and move over to it, and the lock-in is not as real as it used to be in like the monolithic platform, all your data gets sucked up and locked into something. Like, you can actually switch providers more easily now. All right, so give me the product roadmap for Looker. Spill the beans, come on, share the roadmap. What's on the roadmap? Still high level, here's what's specific. So we're still a young company. We definitely have some feature parity catch up to do with the other tools that have been around for 15 years longer than us. So we're trying to learn the things that the tableaus and the micro strategies do well, but do them in the Looker way. So that means enriching our data modeling language. It means enriching the visualization front end. It's to the point that you made earlier about dashboarding being a thing that you just have to do. It's building more bells and whistles on dashboards so people can do the things that they need to do, but it's really focusing and doubling down on the core exploration functionality. Tell me about the company status, employees, funding, new stuff that you're excited about, and what's the bottom line? What's the DNA of Looker? If you had to describe me, Intel's a great example. Moore's Law, very systematic. What's the DNA of Looker employee? So we just want to make customers successful of getting deep into their data. We like being data analysts with our customers, and the product was really built by our founders who just love data. They obsess over exploring data, so it's not static reporting. It's getting into the guts of the data, being a more effective data analyst. We're almost 150 employees, almost 400 customers. We just raised our B about six or 12 months ago, so we're in good shape and we're just kind of, we're trying to do more of exactly what we're doing. We think we're in a unique spot in the market, in database transformation at query time, and we feel like we're the leader in that space, so we're really just trying to tell people about what we're doing, and they are craving that product, so we're kind of getting pulled towards every different direction. Yeah, I think it's a great market too. There's been kind of too much BI for the data analyst out there and not enough reality of, hey, this is the best consuming app. And then right now, companies want people to be productive. Yeah, I mean, when it comes down to it, our core business was venture-back startups. It was the super young, before I was at Looker, I was a company called Hotel Tonight. We were one of the first Looker customers. It was just this young, super fast startup. What we found is that those companies love this in database, super fast, iterative analysis process, but larger companies want that too. It just takes them a little bit longer to make decisions. So the same thing that was true for the Hotel Tonight's and the Uber's is true the further up market that you get. All right, well guys, thanks so much for sharing the insight. Congratulations on the hackathon, VP of product, chief analyst officer from Looker, Joe thanks for coming off HP Vertica, HP software, secure Blue Right Back after the short break live in Boston, Massachusetts, HP big data conference, Blue Right Back.