 Okay, welcome back everyone. We're here live in San Francisco at the Moscone Center. This is theCUBE, our flagship program to go out to the events and expect a signal from the noise. And we're here at the Amazon Summit where all the action's happening. Amazon's rolling out there. They're a little road show. They do this on top of reinvent their big mega show which is where all the action happens in terms of big announcements. But this is about Amazon. They've been doing this since, I remember Amazon started with startups. They roll in, they talk in small groups. Really about learning. It's a learning day. They're really presenting their stuff. They're opening up under the hood. They show the developers and the customers what they got, where they're going. Announcing some key enterprise flavor kind of deals with inforges. You're just seeing Amazon open up the broad-based portfolio of their customer base in the feature-rich cloud that they've been building. Pedal to the Metals, the theme that I see happening here. More features. They're announcing everything under the sun here. Kitchen Sink is coming out. And our next guest is Joey Parsons, the head of operations of a really cool startup in Palo Alto, California, my hometown, called Flipboard. My co-host is Jeff Frick here. I was looking at you, but you're welcome to the queue. Yeah, thanks for having me, guys. I'm excited to be here. So Flipboard is a well-established startup, personally known for its mega-funding. Got some big-time backers, client of Perkins, experienced entrepreneurs. So known for big funding, really which kind of got everyone's attention. But the reality is that's a really good app. A lot of people are using it. The idea of using social media, reading stuff on the web, the magazine of the future is going to be virtual. And consumption of content is a big part. So when you guys built a business in Scratch, we're doing some similar work with CrowdChat, so we kind of know the Amazon story. But obviously, you know, state the obvious. I mean, for us, it's pretty obvious. But like the folks out there looking at the relevance of Amazon, how you guys got into building a business on top of Amazon, and what you continue to innovate on. Give them a view of what that is all about. Yeah, I think it's really important when you're initially kind of building a business, that's your footwear. It's kind of, choose a platform that can grow quickly with you and scale to what you hope your app to be. So when we launched our initial iPad four years ago, we knew that we needed a platform that was going to be able to scale with us. And at that time, building tooling around, scaling out a platform was all day AWS installments. So it really made sense with us to go with a platform where we'd be able to try a lot of things. Like Andy mentioned in the keynote, companies that are like failing fast and being able to experiment a lot. We have a real great opportunity to do that with an Amazon. And that's what we've done at FlipOrdish initially. So talk about what you guys use with Amazon and what you guys build in-house on your own. Yeah, so obviously we're heavy users at EC2. We're in the thousands of instances there. We use CloudFront pretty heavily to sort out our standard content, like our proprietary packaging content to serve directly out of CloudFront. And in terms of like build versus buy decisions, we actually use RDS for MySQL. We're leveraging a lot of the work that they've done there to make that a really seamless experience for us so that we can really focus on scaling out other pieces of our service as opposed to scaling out MySQL. So, and we continue to have, as Amazon continues to release new products, we continue to have kind of like that build versus buy discussion around should we go with like the Amazon provided solution versus rolling with our own. In some cases, we've predated Amazon in a lot of ways where we started using Memcache or Elasticcache-assisted, so it's one of the reasons why we've chosen to continue on with Memcache because changing it midstream isn't always the best idea, but we always continue to reevaluate from both a performance and a cost perspective on what it would take to switch over and make those decisions accordingly. As you guys build more legacy into your application, as you guys stay alive longer and roll out more users, you kind of have to build up these practices. So can you talk about the areas of that around scale in particular? Auto-scaling, Amazon's got some pretty interesting auto-scaling, they got some Elasticcache and they mentioned Memcache. And the promise of DevOps really is about version control, pushing code, infrastructure risk code, as we always say here on the queue. So what's baked and what's not baked? From your perspective, I know Amazon's got some great stuff, but some of this stuff is pretty bleeding edge, right? So you guys have to balance from an off standpoint, tapping into the bleeding edge stuff, but also making sure that it's operationally sound. But what are those areas that you guys look at that are bleeding edge and irrelevant that you guys are kicking the tires on? I think definitely on the bleeding edge side, we're taking a good look at Kinesis. I think from a real-time stream processing standpoint, it makes a lot of sense for Amazon to build that as part of their platform. And we use some open-source tools like Kafka and Storm to accomplish that currently. So continuing to look at that and see if that makes sense for us to switch over to. But that's also a significant piece of our platform and we want to make sure that if we do make the switch, that it will go well. And also things like DynamoDB, we make heavy use of actually HBase and a key value store there. And what we're looking for is something that could potentially replace it, but maybe not, right? So we'll continue to evaluate- DynamoDB is a pretty good place for HBase. Yeah, we'll continue to evaluate and test it and see what things end up. But ultimately, it has to end up in a better experience for our viewers and a faster experience there. And ultimately, that's what we're looking for the most. We'll get that on the record. So I know the geeks that are on crowdchat.net slash AWS Summit. Bring up that question. HBase, which is DynamoDB. I know we ported on, so on the live stream, I want to see some commentary on that from all the HBase database geeks. So sorry, Jeff. Yeah, so talk a little bit about how much of the platform did you guys have initially that you built? Or did you start it on Amazon? And what's this constant trade-off between your own and these AWS services? So we've actually been on AWS from day one. We've never used any other platform for serving out our magazine content or reader content out there. So with that being said, we don't envision a future without Amazon. I think the flexibility that it gives us to be able to try new experience, scale quickly, and kind of meet the needs of how our reader activity goes throughout the day, Amazon makes a lot of sense for us to be able to continue doing that. So talk a little bit about how you're using the cloud and big data and the ability to do A.B. testing to really develop your applications and move your product forward. Yeah, so we've actually built our own internal A.B. testing tool, but in terms of A.B. testing from an infrastructure standpoint, whenever a new instance gets released, we'll actually deploy that with one of our production service roles with a portion of production traffic going to it. And if it holds up and does well, we'll actually switch over immediately on the infrastructure side to use that actual new instance side. And actually during the keynote, when they mentioned the new R3 instances and actually the price drops for C3s and M3s, that really changes the economics of some of the stuff that we're doing. And I'm excited to actually go back and try to see how this changes things. Which leads us into another question. The typical knock on kind of using an AWS service as you grow is yeah, it makes a lot of sense when you're small, but at some point in time, buying is more economical than renting. I think there's definitely some people that have followed that path, but there's also a whole host of other startups and other enterprise companies that are huge, that are continuing to stay on AWS. So I think it's a bet that you make and a calculated bet, but there's lots of history there in terms of companies that have stayed with the cloud and been able to be successful with it. And that's what we're trying to do with football. Right. Adrian Cockroft, a CUBE alumni from Netflix is kind of the post-academic. What about the media bubble right now? Let's just, we'd love to talk about the media bubble because we're a media company. We're 20 full-time employees now, self-funded from scratch. Still not on the Flipboard app. So I want to make sure you guys put us in the Flipboard app with number one in enterprise tech. I don't know why you guys don't have us in there, but I want to get that plug in. The media bubble is about how the consumption's changing. So outside of the overfunding, you guys are heavily funded and it's not just only a bubble on your end, you've got some good backers, they believe in the vision. The reality is consumption is not just about media magazines. It's a user experience change. What you're seeing on the web with social media, for instance, it's a consumption issue, right? So you're seeing the media companies becoming large enterprises because they have to serve up the infrastructure at scale to do some of this consumption issues. So it's a lot like, you guys have to build your own Google in a way. Like you have a lot of content and storing it. So as a media company, there's a lot of bubble growth there, but how does that make you guys think like an enterprise? Are you being specific around some of the operational things you guys do that makes you think about 100% of the time, some of these enterprise features? Yeah, I think from an operational standpoint, our most important thing is for our readers to have a fast, beautiful, always-available experience. You know, and shooting for 100% of time is a personal goal of mine on the operations team to ensure that nobody has a bad reader experience. So in thinking a little bit more like an enterprise with like five nines and things like that, you know, we definitely strive to meet kind of like the same expectations that people have from their core-like life, the core-like life tools and apps they use. We want Flipboard to be one of those and we want to make sure that we provide the uptime to be able to deliver on. Let me ask you a different question then. So can people say, oh, Flipboard's a piece of cake. My kid, he runs the ice cream machine at the freaking local store. He could build that app in a second, spin up Amazon and create a Flipboard competitor. Piece of cake, trivial. Okay, that's what someone might say, they look at like these apps on the iPhone, get it on the iPhone. Talk about how hard it is to do when you guys are doing operationally. I mean, we know, kept some insight into your business, but it's not trivial. Talk about some of the challenges that have gone into the engineering of Flipboard and how you guys have maintained that agile with you guys. I definitely think on the scale side, as I mentioned in the keynote, that we have two big systems. We have real-time interactive systems that serve up real-time content to our millions of readers. And we also have our data importing platform that's going out to our publisher content, RSS feeds and real-time streams of social activity. So if you think about it, if you're looking at something that came in from your social network that you recently liked there and you open it up in Flipboard, you'll see it right there that you actually liked it. Being able to have that kind of ability to be able to show your recent engagement is actually a pretty tough problem to solve if you look at the scale of how many users we have and how many people are actually using the system. A lot of I.O. too, a lot of I.O. too. Definitely a lot of I.O. You're building up data stores and being able to retrieve information randomly from those users. Do you use a lot of Node.js on that stuff? Actually, our web front end is actually written in Node.js. But obviously, the other clients are written in their native languages on Android and iOS. All right, so let me ask the harder question, which is, if you didn't have Amazon, what would your staff look like? Do you need to pay a picture of, maybe you're not old like me, but I actually remember stacking in a rack and servers that really kind of sucks. But what would their life be like, just staff-wise? Just what would be the picture? Without Amazon? I think it would easily be past 15 people. Just to manage it. Easily fast to be like, yeah. From people managing the network to like the racking and stacking like you mentioned, and even people that are dealing with buying stuff, like the procurement of stuff and dealing with the provisioning and shipping and getting it into your cage and dealing with those relationships with all the different providers that you have to deal with, easily at 15, and we've been able to deal with just three. So scaling to 100 million readers with just a full-time off-staff of just three people. We're not going to hold you as a guesstimation. So it's about 15, yeah, it's probably around 20, maybe 20, maybe 25, it could be larger. If you think about using different availability zones in different regions and having to have staff in each one of those regions to be able to service that, it's a pretty tough thing. And Joe is a mild mannered guy and he's friendly and smiley, but if you look at his LinkedIn page, John, I wanted to check it and do some of your research but they're right at the top, nothing short of kicking ass is acceptable. So this is not a trivial pursuit to do Flipboard. It's not a trivial pursuit to start media startup. So, you know, be careful, right? Definitely, yeah. He's not going to be a pushover. Right. Okay, Joey Parsons, thanks for coming on the Cube Flipboard. Guys, check out Flipboard. If you don't already know about it, most people do know about it. It's a really hot app. It was really one of the first apps that came out that was really focused in on the elegance of the beauty of the product, which is what I always was attracted to. It was the local hometown favorite, Palo Alto. This is the Cube. We'll be right back with our next guest. We're live in San Francisco at the Amazon Summit, Amazon Web Services Summit here in San Francisco. This is the Cube. We'll be right back with our next guest.