 Okay, we're back here live inside the Cube of SiliconANGLE.tv's exclusive coverage of the HBase conference. This is where all the alpha geeks and the ninjas and the pirates, they're all here building the next generation. I call the holy trinity of the Hadoop ecosystem. HDFS, HBase, and MapReduce, all kind of coming together to create highly scalable marketplace HBase conferences, changing the game on the data warehousing, business intelligence, a real-time element. This is where the action's happening and we're here live and I'm here with Danny Ryan, who's the data ninja at Riot Games. Danny, ex-Cloud era, we knew each other at Cloud era. You did your thesis on a project that we worked on together. That we've had the chief data scientist, Michelle Bailey, for VDP Finder. Great contributor to the community. Great guy, welcome to the Cube. Thanks John, thanks for a lovely introduction. Yeah, I'm pretty excited to see where HBase is going. Been using HBase for quite some time. It's been 2.5 years since I started using HBase. It's really, really a good data storage system, good for real-time applications and good for big data. Yeah, one of my thesis project was to do something over the social media analytics. It's very hard, there are a lot of people doing that and many get it right and many get it in a different way. There's a lot of things that are happening and one of the talk today was, again, like using HBase for graphs by some company called FullContact.com. So it's, again, it's echoing the same sentiment. You can have graphs in HBase, you can have social media in HBase, you can aggregate real-time on HBase. HBase, on top of that, is a kind of layer which allows you to do a lot of cool stuff, which is not possible before. So let's take your social media analytics project, get it for your thesis. And then we'll talk about Ryan games in a quick second. So what did you hear today that gives you new insight into that kind of project around social media analytics? What was new? What was new was, we can now store terabytes of information just on a huge string budget. That is, a couple of commodity hardware, put a free open source software on top of it, opens a stack, and here, boom you go, you have eight to nine terabytes of data, which is accessible, which is completely indexable, and you can do low-latency QLU access, which is really cool, which is not possible a decade ago. You have to go for enterprise-grade software systems, which will cost you a lot of millions of dollars, or sometimes, like hundreds and thousands of dollars. So there was a lot of impediments to using those systems. So HBase actually simplified a lot of things for us. If you are in the community, you get to know how the system works. And of course, the community is there to help you with all sorts of schema design, answer your questions, and it's great, and we grow together, like, HBase improved, and we started improving with it. Like, now we can store like, hundreds of bytes per second in a real-time fashion. So what's going on at Riot Games right now? So I'll see you at Cloud Era, did your master's internship at Cloud Era, and then graduated and went to Riot Games, heading up the data platform there, architecting that out, what is going on there? Tell us a little bit about the environment, and the architecture, and kind of the business metrics and zillions of users. What's the numbers? Yeah, so I can share the public numbers with you. So the public numbers were raised like last November. We have grown since then, but the public numbers are like 32 million registered gaming. 30 million. 32 million registered. 32 million registered gaming. Registered gamers across the planet, 145 countries, four million unique logins every day. 10.5 million gaming hours every day. So that is different than just visiting a website. It's very hardcore. It's the average user- That's hyperscale. Scale is huge, like lots of gaming servers. I can say like thousands of them, and we generate a lot of data, of course. So that's a huge deal. So we are still like, we will be using HedgePace, and we are like doing a lot of proof of concepts for using HedgePace. We are finding the use cases for it. And I'm giving a talk at the Hadoop Summit regarding the use cases of Hadoop at Riot Games. That's a good plug. So Hadoop Summit is June 12th and 13th, or 13th and 14th here in Santa Clara in the Bay Area. If you're really interested, there'll be more tech talks there. So you'll be speaking there. I will be speaking there regarding how we use Hadoop and Hive to leverage the data which Riot Games produces. So that's something which we have as a production called the system. You're in production with it? Yeah, we are in production with Hadoop and Hive stuff. It's used for backend analytics. Is H-Pace in production? No. So we are looking for use cases. We know that H-Pace is really awesome for real-time analytics. What does near real-time mean? Near real-time means seconds. Not nanoseconds. No, never. No. You can go to sub milliseconds at max with H-Pace. Nanoseconds, Hadoop X and Millie. That's RAM. That's called Flash Power. Yeah, that's like flash drives and SSDs. But of course, with the scale which we talk, like web scale, near real-time is good. People don't expect your data to be aggregated in nanoseconds. We are not running banks, right? No, no, no. Well, you can actually run some front-end stuff kind of aload pages and aload some images in there and bring it in within a second. Yeah, but you don't aload a million dollars if there is a millisecond interval. Just like financial companies. Gaming environment, multiplayer gaming, it's really, seconds do matter. Transaction starts in virtual currency. Yeah, yeah, yeah. So we have more thing about the analytics use cases which is not transaction-oriented. H-Pace, even as of now, it's just a guarantee. Good for analytics. Yeah, good for analytics, but not good for transactions. What are you using for the transactions? For transactions, people use the regular stuff. So I'm not like, I'm not on that stack. It's OLTP stuff. Yeah, regular OLTP stuff. I'm not on that stack. I'm more about the big data stuff, how to use the logs, how to use like a ton of data-write generators, how to mine interesting patterns, those kind of stuff. Yeah, yeah. Yeah, and same with like every big data project. You get, you design a schema, you have it right, and you run a bunch of mapper diesel gardens. Yeah, you know, you're very entrepreneurial and get to know you over the past couple years. It's exciting time, and you know, Riot Games is one of those startups that is looking really good. Yeah, yeah, yeah, yeah, yeah, yeah. As well as what's going on in the community. I know you're very active with your peers, and that's great. Sure, sure, sure, sure. I'm looking forward to like every bit of opportunity available here. I'm glad that my master's thesis actually is good in terms of solving a real business world problem. Very glad to know about it. So obviously you were, you know, it's been kicked around and it's really not really that big of a public secret, but you know, you were actively involved in helping me get VVP Finder off the ground when you were at Cloudera transitioning to get your master's degree. I showed you the current product. You haven't seen it in a while, so what do you think of the current situation of that product? I mean, Michelle's obviously doing some good work in the field with customers and the field trials and the engineering team that you know through the community is doing some good work. What do you think of the progress to date? I think it's a very awesome progress. It's something which is not done so far. I've been getting people in real time. It's a win, right? So I'm not sure many of you start. Have you seen this anywhere else? Yeah, of course. No, no, no, no, no, I've not seen. Have you seen anyone doing what we're doing? Yeah, no. The thing is I would love to get it patented as soon as possible before talking on camera. Don't just, we are patenting it, so. Oh, we are patenting it, no worries. I'll get you to get a little byline, but you know our engineers have been doing some good work. Michelle's doing some good work out there and the field trials are very successful right now. From a technical perspective, H base as a batch now real time, some of the things that we've done, I showed you, is pretty compelling. So just, how far are we ahead? Are we ahead? Yeah, we're ahead. Let me tell you this pyramid. So around like 2000, Yahoo was a pinnacle company because they showed web pages. Then Google came and you can search like millions of web pages and then Facebook came and you can see millions of people. And this pyramid simply grows. Like there should be something to do with people, right? People are on the internet. People are everywhere. So something to search, something to like. Yeah. Trigger and like group people, those kind of stuff. Your vision is right on, you know. You're present, you created that product with me. So your vision is fantastic. Yeah, I agree. Yeah, I don't know. People, it's a people web. I mean it's about people. It's about people, it's going to be about people and there are multiple social networks now. There's LinkedIn, there's Facebook, there's Instagram, there's Twitter. So you can see now the world is having like multiple aggregates of people. So there should be something on top of it. So you go and go on building the pyramid. It should be great to get that an interesting analytic layer. Yep, analytical layer. That's HBase, so that's kind of what we're doing. Yeah, that's right. Very exciting, I wish I could talk more about it. Danny, I'm glad that you were part of it. VDP finder, Michelle's leading the charge there. It's a real big success story for HBase. And you know, Mike Olson was really impressed when I showed it to him and all the Cloudera guys. And I'm proud to say we're using Cloudera manager, so. Yeah, that's something which actually simplified the adoption. Hey, just like install it in a... Cloudera, save months. Months of savings. Yeah, and you don't have to like automate any stuff. It's cool, it's a component. So what do you think about this conference? What's your wrap up of the conference? This conference is the first HBase conference of many to happen. And this is very exciting for the first conference. There's so many people, like it's just like one product and we have like many people working. So... It's Christoph Priscilla, he was on. Christoph was on talking about his product, another HBase cohort. What do you think about Weeby Datum, some of the stuff that they're doing? Yeah, Weeby Datum is an exceptional product. This is a schema layer on top of HBase and it's very generic. It can be used for social media, it can be used for any kind of like entity-centered data. So it's people are... He's a cool guy too. So yeah, people will be designing multiple focused stuff, similar to Weeby Data. It's the fact that Weeby Data exists actually points to us. Yeah, there's something you have to... There's an application middle layer opportunity. That's right, there's an opportunity there. That's what I suppose, like yeah. So every application requires a very specific schema. This is what you understand. HBase is not like MySQL, you just insert, you just install it and you have tables bloom, you can have your app. It's not like lamp stack, it's not ready. It's not, it has to be tailored. It's like a tailored suit, which... HBase is a tailored suit, I like that. Yeah, HBase is a tailored suit. HBase is like high-end custom suit. MySQL is off the rack. I'm off the rack, it's just like... I'm going to use that line, nice, that's good. Or like a sports car. These be tailored. Yeah, just be tailored for you and for your use case. And there's a ton of opportunity there. Like every application needs a schema design, which is not very intuitive. There are many stuff which you need to understand. When you use HBase, the rookie design, the single most important factor that determines what it is. So what needs to happen? Will there be a common layer? I mean, someone has to fill that void. Yeah, that's right. Is it Webidata? Is it VDP? What is it? Every vertical line, every segment of business will need its own use case for HBase and I guess there are going to be multiple players. So one company will be focused on designing something for a particular CRM application on top of HBase, maybe. And some other company may be focused on social media application. Because you have to tailor it. You cannot just use it. So everyone... So the ease of use, the tools need to be developed. Yeah, that's good. So that's a huge gap. So I think there are going to be multiple players and it's going to be awesome time ahead. Okay, so what's next for Riot Games? Riot Games is expanding worldwide. It's a global phenomenon and it's taking off the world and people love our games. How do you like living in Santa Monica? Oh wow, that's a good question. Versus Silicon Valley. Come on, come clean. Tell us, tell us. Santa Monica is all about beautiful nature in terms of climate and everything else. People. People, stylish people. You often meet people who are really awesome. Here, of course, you meet geeks on the road. There you run into beautiful girls. There it is. Beautiful people. Okay, Danny Ryan with Riot Games. Good friend and co-founder of VDP. Just on a project, he continues to be a strategic advisor to VDP Finder, that Michelle's running. Great stuff that we've been part of at Silicon Valley, SiliconANGLE Labs. Congratulations, great to see you. We'll be right back with a wrap up after this short break and Michelle Bailey and myself will wrap up the HBase conference right after this break.