(upbeat techno music) >> Narrator: From New York, it's theCUBE, Covering Escape/19. >> Okay, welcome back to theCUBE coverage here in New York City for the first inaugural Multi-Cloud Conference called Escape/2019 as in gathering of industry thought leaders, experts, entrepreneurs, engineers, really having substantive conversations around what multi-cloud is, what it's going to look like, what are some of the thing, technical and business opportunities around that, really small intimate conference. Again first inaugural conference. I'm here with my next guest to talk about that Chris Smith, Vice President of Engineering, on Data Science at Ticketmaster. Chris, thanks for coming on. >> Thank you very much Don. >> Appreciate taking the time. >> Glad to talk to you. >> Practitioner out there, you know, we all go scar tissue. >> Yes we do. >> If you don't have scar tissue, if you're not breaking things and then the learning from it then you're not advancing. But sometimes you don't want to step too far forward right? >> Yep, yep. >> Can you get back it's like you know. So you guys have a great experience. Legacy business, I remember buying tickets when I was going to conference back in the day when I was in, you know, in college. >> Yep. >> Buy it at Ticketmaster. >> That's right, that was Ticketmaster then, Ticketmaster now. >> Now it's lot of online provisioning of all direct to consumer. So you guys are a journey, tell the story. >> Well certainly, the company Ticketmaster, has had an incredibly long journey, starting back our first concert was Electric Light Orchestra which kind of like puts that in in context. >> (laughs) I was in eighth grade, '79. >> Yeah, yeah that was back at ASU. And even then we were a very innovative technology company we were making ticketing platforms that performed better, got more capacity out of the hardware than anybody else could do, anything close to that. We were really pioneered that idea of the what was at the time called the electronic ticket. Which was the idea that, you know, you could go to any store that was selling tickets for an event and the same inventory would be available at each store instead of the old model of a bunch of tickets getting sent out to each place >> That was bad-ass back in the day. >> That was really cutting edge and we've been evolving ever since then for 40 years. We were also very early onto the web scene. We were selling tickets online before anybody else was and before most people were selling anything online really to a degree. So we've been pioneers in a lot of areas, we see ourselves as the technology partner for the live events business. That's really what we are. And as a consequence, we're always sitting on that edge right? Trying to innovate and move to new opportunities but at the same time trying to provide that quality of experience at scale. >> Yeah. >> That is so critical to the business. >> And there's a big business so it's not like it's your nimble start up but you got to be agile. What are the learnings? Take us through the cloud learnings as you guys pioneered and started to go into that pioneering mode which was okay, you don't have to be a rocket scientist to figure out what a cloud's going to do. So you guys probably said hey, we got to go look at this, let's go pioneer our impact, take us through that what happened? >> Yeah absolutely, and I think there's two interesting contexts that started that conversation right? One was we're one of the few online businesses that launches a denial of services attack against itself on a regular basis, basically every day, right? And so we have traffic patterns that are unusual even for a typical e-commerce site where we might see loads that are a hundred x, you know beginning of a Taylor Swift on sale. There's going to be traffic like no one's business. And then when all her tickets are sold, there's not going to be nearly as much traffic right? And so that is the nature of our business and cloud is very attractive for its elastic capacity. When we were running on prim, we have to provide all that capacity all the time, just to have it for that one peak moment that might literally be the highest traffic level we see all year, right? So that drew a lot of the interest in looking at the cloud in the first place. And then the other aspect was we'd been working on, you know we'd been running on prim for nearly 40 years at the time and there is a lot of technical debt that had accumulated in the system at that point. And so, there was an interest in maybe potentially being able to leverage cloud vendors' infrastructure, and migrate systems onto that and then sort of declare bankruptcy on some of that technical debt rather than trying to pay it off. And so that, those were the two thoughts that were driving that conversation. I think we got really excited by the possibility and we committed really heavily to the idea of a strategy of just moving aggressively into the cloud as fast as we possibly could. And we knew that in the process, that we would be breaking some things, we'd be you know discovering some challenges et cetera, and that's definitely what happened, right? >> (laughs) What was the big learning? >> I think the biggest learning was that, you know, we had been developing systems for decades literally, with our on prim environment and so the systems were actually very well tuned for that on prim environment and that on prim environment was very well tuned for them. >> Yeah, yeah exactly. >> And it clouds use-- >> On all levels, hardware, software. >> Yeah, all the way through 'cause it's a fully integrated, vertically integrated solution. We build a lot of this stuff custom ourselves. >> John: Yeah, and we would decompose all that. >> And so it was very difficult to migrate some parts of that to the cloud and more importantly we're pretty smart guys, we can figure out how to move stuff into the cloud. But then to do it in a cost effective manner. Required in a lot of cases, really dramatically changing the design and architecture even of the software at a pretty fundamental level that you just can't do overnight. And so ironically, you know, the technical debt that we had in our infrastructure didn't seem quite so huge once you start thinking about the technical debt of the entire stack, right? And so then we realized that we could be much more strategic about how we went after our cloud strategy and that's kind of where we are now. Where we are being smart about, there's a lot of new products that are being developed, that, you know, we can build from the get go with the idea of them being designed for the cloud. >> Cloud native. >> Exactly, so we have a lot of stuff like that, that's just being built, in fact, the bulk of our website now when you go to visit it as a consumer, the bulk of that is running in the cloud right now. But, there are some really critical systems that are core to that experience, that are still running on prim. >> So you guys had to essentially re-architect the operating environment to take into account hybrid operating. >> Yes. >> Decoupling the critical systems that can't be tampered with, maybe put some containers of Kubernetes move some services around. But for the most part treat Cloud Native as Cloud Native, Greenfield apps and nurture-- >> Yeah but there's also refactoring opportunities. So there's a lot of opportunities where you need to go in and change the product anyway and that can be an opportunity to make things a lot more cloud friendly and better take advantage of the capabilities that the cloud has, so it's actually a mix of both. >> Give an example of a good opportunity to refactory, 'cause this comes up a lot in my CUBE interviews. Like okay, 'cause it's all opportunity, opportunistic, but what are the characteristics for a great refactoring opportunity the tune up? >> So a lot of times when you want to refactor really what you want to do is take a set of capabilities that you may have in a much larger system and pull 'em out and manipulate them and play around with them and do things differently. So, our ticket purchasing process we're constantly looking at tweaking the process. Now the core pieces of it remain the same right? But we might want to change the experience and provide something more innovative that's different from what people used to do. And so one of the areas we're working on for this as an example is reserve-less checkout. Where you just buy the ticket without ever actually reserving the seat. That's a very small minor change in the flow, but to make that really work you have to pull out the pieces of the system anyway right? And grab, say I want these four pieces to rearrange differently, so that's a great refactoring opportunity. You can make all those pieces, what we actually did is we've made those pieces into lambdas that are sitting in AWS, they're basically not running most of the time which is great. >> Yeah. (laughs) >> Really cheap when it's not running right? >> Yeah, exactly. >> Very efficient. But then when we need them they run very efficiently and more importantly we can now manipulate the order of operations for this stuff. So breaking things out into those composable parts whenever you know you need to do that anyway, it's a great opportunity to change it. >> So great for work flow refactoring there. >> Absolutely. >> Final question for you, I know we got to break for lunch, but, then really appreciate you coming and sharing your insight. >> Absolutely. >> As a pioneer in data science and data you got machine learning certainly is the engine of AI. AI gets math and cognition are kind of coming into it. Learning machines, deep learning, bla bla bla, what's your, in your opinion, what are some pioneering areas that are ripe pioneering grounds to dig into in data science and data? When you think about CloudScale, Hybrid and just, in general what are the ripe opportunities for people to pioneer in daily. What's the next frontier in your mind? >> So I think the trend right now that's maybe not the frontier, but it's now where the main shift is, is to moving into what I would call real time learning, right? Where you're doing refactor, reinforcement learning, or online learning of some form. Where you're literally, the data's arriving in real time, transforming your model in real time, learning in real time, that's key to our strategy and it's very very common. But I think in terms of where the frontiers are it's actually kind of everywhere, in the sense that the name of the game is the cost of doing that work is getting lower and lower. You know, data's getting cheaper, computes' getting cheaper, and also the products for doing it are getting more productized, so you need less expertise and you can deploy them more quickly. So what you want to look at is businesses that are traditionally been too low margin right? To apply machine learning to but have large scale, right? Which is like the commodity, everything in that's commoditized, right? Now there's an opportunity to, there's the cost have gone so low-- >> To squeeze insight out of those areas. >> That you can now optimize that small margin and get value from it with you know, otherwise like 10 years ago it would have been so costly to build a machine learning infrastructure for it. You would've lost more money than you would've gained. >> So you could, what your saying is, these areas that were not attractive because of cost in the past, that have large scale, there's penetration opportunities to create value and insight that could-- >> Absolutely. >> Bring in new franchises and new capabilities. >> And that's why I think you know the Andreessen's software's eating the world thing, that's what that's really about is as those costs get lower, as the ability to deploy gets easier, suddenly businesses that before didn't make any sense to invest in this way, they totally make sense and in fact there's huge opportunities to completely transform the landscape by getting in. >> Chris you're a man of our world, we love you, thank you for coming on theCUBE. >> Thank you so much. >> That's great insight. >> Look at this we're getting insider on the future of data, which I believe everything that he just said is totally relevant. You're an entrepreneur out there, you can attack big markets and get in there with a position with great IP, great intellectual property, again this is the modern world of computer science. >> It is. >> Don't ya think? >> It absolutely is. >> This is the benefit of scale and cloud. >> Absolutely. >> I wish I was 20 something years old again. (laughs) We've been through the ringer. >> Yes. >> Chris, thanks for coming on. Keep coverage here in New York for the first inaugural conference, Escape/2019, I'm John Furrier here, thanks for watching. (upbeat techno music)