 Okay, we're back here live in Las Vegas. This is theCUBE, our exclusive coverage of Amazon Web Services, re-invent conference. We're on the ground, I'm John Furrier, the founder of SiliconANGLE, I'm Joe Mai Co-host. As always, Dave Vellante for the day two wrap up. They're literally shutting the lights down here and on the ground floor. People moving out to their next level of events, which is all social networking activities that we will not be going out tonight. A lot of great action. Dave, let's do a wrap up of day two. I'll start and then I want to get your perspective. I just think overall, really love the awkwardness of Amazon's growth. It's like a pre-teen, going from middle school to high school, they're awkward, they're getting stronger, they're growing, and you can see some of the rawness of Amazon, the people we talk to are just so kick-ass, technical. They don't have the polish yet of the big machinery. And that's really my big takeaway. And you know what, I like it. I think that's really, really good. James Hamilton came on. This guy's a distinguished engineer. Old school been with Microsoft and for that IBM. Intoxicated by the prospects of Amazon, technically and also from a business impact. But that rawness is really refreshing. And I think the, hey, here are our cards. We're putting it on the table. Finally we see Amazon put some cards on the table. And guess what it says? We're coming to the enterprise. That's kind of the take, my take. What do you think? I agree with you, John. I really like the way Martin from Eucalyptus, Eucalyptus put it, the CEO of Eucalyptus. He said, economies have always shown over history that some people will rent, some people will own. And I agree with that. I think that sums up what we're seeing here. There's clearly a schism. And you're going to see private cloud and you're going to see public cloud. I think the winds at the back of public cloud. I loved also the second real big takeaway for me today was to underscore that schism, that bifurcation of opinion. Andy Jassy put up a slide and it was 80%, it was the legacy vendor's view of cloud. And 80% of the slide, the pie, pie chart, 80% of the pie was on-premise, 20% was cloud. And then he said, our view of the world, just the opposite, 20% on-premise, 80% cloud. And that's the big battle. Now having said all that, I think it's very clear to me, John, while some of these harder to move applications are moving and you see companies like GE and Pfizer and Shell and all these list of companies that are doing traditional enterprise companies that are moving to the cloud, the fact is the sweet spot of the cloud generally and Amazon's cloud are greenfield applications and startups. So John, I want to talk to you about why. I want to talk about the experience that you've had in developing CrowdChat. We've been hearing about CrowdChat and CrowdSpots for quite some time now. You are the perfect case study in my mind because you started off with bare metal in a colo with HBase in a key value store and you and your team decided to move to Amazon. I want you to take us back to that point, what was it, a year and a half ago? Actually, three years ago when you started the project and then take us through how you ended up on AWS. Well, Dave, that's a good question. Thanks for teaming that up. I'll do my best I can off the top of my head but to get some context, we built an idea that we could use data. This is back in 2009 and then my experience at Cloudera 2010, 2010 and then 2011 really became clear that the database market was shifting as James Hamilton pointed out. So we basically conceived of a notion of using data to get access to information around people and their behaviors on Twitter, for instance. So we built a very simple system that basically takes Twitter data and sucks it into an unstructured database. And without going into all the details, it was really easy to do. We rolled it out on Hadoop and then obviously HBase, Danny Ryan, the co-founder also from Cloudera, we basically got HBase up and running very quickly because of our experience with Cloudera and HBase. And we built Solar Search on top of HBase a year and a half before Cloudera shipped their version. So we did a lot of coolness. However, at the scale of the tweets we were getting, we couldn't handle the load required, a little bit too much cost of being a bootstrap startup because it's a startup, we couldn't afford. I mean, we could have afforded it, but I didn't want to pay for seven more micro boxes or cheap hardware. But at first, you chose not to go to Amazon because on paper it looked expensive, right? You said, all right, let's prototype the thing on one box in a closet. Basically the POC was done on Hadoop and HBase, right? So that was on one box, so yeah, it was multiple machines. And then you realized this thing's never going to scale. It's too complicated, too many people. It was basically a 2U box with four machines in it, maxed out, I think it was like $8,000, whatever, test machine basically, it's on the network. So when we started looking at the scale piece, it just wasn't right for us, budget-wise, and also labor-intensive. So we moved everything to Amazon. The game changed significantly. We now run all full stack on Amazon, we're handling 75 million unique Twitter users, we're monitoring every day using machine learning and classification on their interests, cross that, connecting that with their social graphs, and doing that in real time, 75 million people, and connecting the data between 75 million people on Twitter across multiple 300 verticals and underneath 10 million hashtags. So technically the challenge was, how do we do all this? And we've done it, we've run it all on us. That's 75 million unique people, fully classified by all their interests, fully searchable within less than a second latency, and running completely across now with social network data across 300 verticals and 10 million hashtags. And you're running algorithms against that data and machine learning to essentially segment it. Yeah, so some IP, a lot of IP involved. A lot of machine learning, a lot of intelligence there, custom software, we have something called Affinity Rank, which is potentially a proprietary protocol that goes out and basically assigns every individual an affinity score based on any given context in real time, meaning that we can search on an individual and saying, what's this affinity score? And then just throw a concept at it, say cloud and we'd have a number, you say developer, we'd throw a number. So we can essentially rank people in real time based on the data set that we have. That's pretty compelling. So when you went to AWS, you didn't just flick a switch and go to AWS, it took some development. Talk about that a little bit. So we have all the goodness of, and Danny could probably do a better job explaining it, but we have all the goodness of Amazon. We got EC2, RDS, all that great stuff. But the real key was full text solar cloud on ADS machine spinning up using SQS, simple query service, huge, huge queuing capabilities now. We don't have to do any work. All we just push code and everything happens. We run our entire CrowdChat app using Node and Elastic Beanstalk. It was very well integrated into Git. Easy to maintain, multiple versions of applications. A lot of great stuff. We ended up building our own Redis cluster and then Amazon just recently on September 4th announced their own. So we had that up and running before then. But it took some time to develop that infrastructure. I don't know how long. I mean, it took a couple of weeks. I mean, it's not significant, not years. I mean, we kept on weeks or months, right? Well, we had functionality up and running immediately. And then it was essentially iteration. And I'm oversimplifying it, but we're constantly tweaking it. But it really is a dream day. We now have a fully scalable back end from a hardware standpoint and software. So all of our IP is in software, okay? So the entire scale piece really breaks down into two categories. Writing software that scales to the levels that we infinite scale basically in our mind. Scaling software-wise, that's architecture, that's how the code's written. And then actually having software, hardware infrastructure that scales. That's where Amazon comes in. So Amazon, we can handle floods of traffic from Superbowl, Oscars, any kind of social data at massive volumes with four guys. I mean, that's the kind of level of scale. So Amazon's only as good as our software, which is actually really, really good. So we now have a full system. Now we're going to work on the front end, the Node.js stuff and then the user experience. But CrowdChat's been a great success. So what happens when like, let's say we're doing a CrowdChat on the Superbowl and a gazillion people came in. How quickly would you be able to scale up spin up servers and accommodate that demand? Well, we have the auto-scaling. So it automatically kicks in and we have policy that says, okay, based upon certain loads, we just scale up more machines. And we just scale up and then scale them seamlessly. Seamless scaling, yeah. And that's all programmable via dashboard. So I tweeted out today one of the testimonials at the beginning of the re-invent keynotes. One of the customers said, I would literally break out in tears if they took AWS away from me. I feel like your team feels the same way, Danny, if you're watching, I wonder if you confirm that. Is that how you feel? I don't know, we could port to bare metal, just be costly, we'd have to get paid for machines. So basically with bare metal, ultimately all startups that are on Amazon have in the past moved to hosting their own. I'm not sure if it's been a control issue. Like you said, some people buy and some people rent. Some people go all pure cloud like Amazon showcasing. So I think it's getting seamless. I think you'll start to see people use all cloud. Well, we'll certainly use all cloud. But when you do your own bare metal, you have to pay a lot of upfront sunk costs. And you can still kind of scale up with bare metal, but that's machines, network, and people. You need sysadmin, you need a database guy. So when stuff breaks, you need management support. You got to come in there, reboot the servers, run the clusters. I mean, that's in the hundreds of thousands of dollars. Significant cash and using Amazon, all the labor costs is in code. And the rest is all variable. So you only pay when you're successful. So that's really all you got to be mindful of with Amazon is making sure that your success technically doesn't drive up a huge EC2 bill or S3 bill. That's the only risk you're going to take. And some startups went out of business. One high profile startup flamed out in San Francisco, picture started because their bill was massive and they had no business model. So that's all you got to watch out for. So what's your vision for CrowdChat and CrowdSpots? Well, we don't know yet. It's kind of evolving, it's taking the life of its own. But one thing that Danny and I are excited about with CrowdSpots and CrowdChat is, CrowdSpots, or we used to call it VFinder, is an incredibly rich technology platform that is just amazing. It's pretty badass. Probably no one understands it, right? So we rolled out CrowdChat as a way to build engagement, a chat room on Twitter, LinkedIn, and Facebook to build engagement for the audience. That, everyone understands. So that has been the epiphany, is that the CrowdChat application is a lot like Reddit, AMA on steroids. If you're familiar with Reddit, ask me anything. But it's all legit, you got to sign in with OAuth. And that's been amazing. But the vision really is, Dave, is connecting people and that using tools that are targeted, we have the ability to automate, dial up people in real time on any query. And flash mob, specific unique profiles. And put them in a chat room, in a conversation. And I think that's honestly where our passion is, is to enable people to connect with each other. And save people time, and getting that done. So that's our goal. And then building all that network data underneath it, it's got a network effect. So CrowdChat is exciting because it works well. People love it. It provides engagement value. And you'll see a lot more of that. Well, so thanks for doing that kind of on the spot. So the reason I wanted to talk about that, John, is because I do think it underscores the affinity that startups will have or have for AWS and how AWS is a perfect platform for startups. Now having said that, a lot of people say, well, the great thing about AWS is that startups now have the same access to resources that large companies have. And that's true. But you know what, I'll make another observation, just walking around this event. Large companies that have been stuck in the mud because of IT for years, also now can tap that innovation. And I think that's what's happening now. And that's why this is a trillion dollar opportunity because it's not just startups, it's anybody, it's divisions within corporations, it's Skunkworks, they can now innovate. And Andy Jassy said, you know, I get a lot of times, I get the question, how much money you save? He said the real question you should be asking is, how can I innovate faster within my organization? I think that's what's happening with large companies. You know, Dave, just to kind of go back to CrowdCheck for one minute, I'll just add, because that's a great point, is that it's all about the people. And one thing that people aren't talking about is, we build CrowdCheck because it focuses on people. And one thing about Amazon's doing is that the people who are changing the game, the transformers in these markets, these change agents, are going to create massive opportunities by being creative. So don't underestimate people out there, the creativity we heard from Jason about Cycle. He did stuff that you can never imagine before. So I believe that the creativity of the people will drive everything. And that's going to be the game changer. So I think Cloud will have a swift adoption over the next couple of innings. We're top of the first inning. She's a baseball metaphor. I think it's going to be a nice, good mid-inning growth. You're going to see certainly a lot of adoption. So I'm bullish on it. I think the mindset, the business value is there. The investment is doable. One in the sense it's lower cost than anything else. The only issues are those little table stakes. Compliance, security, the TSA at the airport, getting through security. So I think whoever can build those table stakes features into that Cloud model, whether it's eucalyptus with private Cloud, you just open source, or not, that's the winner in my mind. I think that's going to be whoever can put that together will win everything. Awesome, John. Well, it was a great day today. We're going to be going wall to wall tomorrow. Check out siliconangle.com for all the blogs associated with the show. So siliconangle.tv, that's what we'll be broadcasting live tomorrow, youtube.com slash siliconangle. And also check out wikibon.org for all the research. And we've got a new innovation up. What's the website? The cube.siliconangle. The cube.siliconangle.tv. The cube.siliconangle.tv. And check out the new innovation that we have, essentially a table of contents, if you will, and all the videos indexed, all searchable. New innovation from the cube team. Great job. So really appreciate everybody watching. All right, so we'll see you tomorrow. Stay with us tomorrow all day live coverage. Good night.