 Live from San Francisco, it's theCUBE. Covering DockerCon 18. Brought to you by Docker and its ecosystem partners. Welcome back to theCUBE's coverage of DockerCon 2018. I'm Lisa Martin in San Francisco with John Scheuer. And we're excited to welcome to theCUBE for the first time Ed sued the BPF product and product marketing at Mesesphere. Great to have you on, Ed. Thank you, pleasure to be here. So, Mesesphere, tell us about you guys. What do you do? Why are you at DockerCon? Yeah, so Mesesphere is a hybrid cloud software platform. We basically enable you to very easily adopt all types of new cloud native technologies like Docker, Kubernetes, Spark, all the things that you think about to build these world changing applications. We automate that for you to run on hybrid cloud infrastructures. Nice. So maybe you could break it down a little bit more. I mean, I know people can sometimes get confused Mesesphere the company, Mesesphere the project, DCOS the product, and then Kubernetes and then we're here at DockerCon. Maybe kind of untangle some of those things a little bit. Maybe I'll go in chronological order. So Apache Mesos was actually created way back I think around 2010 as a project to figure out if you had to rebuild Google's proprietary architecture for hyperscale computing, what might that look like? So that became this project called Apache Mesos. Later on at companies like Twitter, Airbnb it started being used to solve some real challenges around scalability and performance. Arguably without Mesos as a technology, I don't think Twitter would probably exist today because Twitter used to crash a lot. You guys remember that? You got the fail-wail picture and all that stuff. Apparently Justin Bieber used to crash Twitter, right? And Mesos became the part of the solution. Now you fast forward a few years later, containerization really caught on, right? And then Docker became a game changer in terms of making sure people start using and adopting container technologies, really popularized containerization. And of course, Kubernetes later came along as a way to orchestrate the operations of these containers. Now where Mesosphere fits in is our platform is actually below a container orchestrator, right? So Kubernetes is actually the fifth container orchestrator to run on Mesos. There's earlier ones like Netflix, I think Twitter themselves, there's different types of container orchestration tools and then Kubernetes became the most recent and frankly most popular container orchestration tool and Mesosphere enables customers to really get one turnkey installation and operations of that technology. You mentioned Netflix and thinking, it powers a lot of our lives. But I'm thinking about IoT data-driven applications like that. How does Mesosphere help power IoT and those data-driven applications? So any IoT application probably needs at least three major sets of capabilities. The first is you have to ingest tons of data. If you're a connected car or your home appliance company, there's a lot of data coming in from all these internet-connected devices. You need a way to ingest all that data without losing any of it and making sure you can be responsive. You also want to be able to analyze that data, right? So tools like Spark and other things that become very important. You also need to be able to host an application or service and Kubernetes becoming really the most popular way to serve these applications. The last and by far I think most important piece for IoT use cases is the concept of hybrid cloud and edge computing. So Mesosphere, we have many connected car companies that are doing connected car or self-driving car projects are actually working with us. And the reason for this is we provide consistency for running containers with Kubernetes or data services like Spark and Kafka on a really elastic infrastructure that can be on a data center, on AWS, on Google or beneath a cell tower or even a cruise ship. Those are all actual use cases. We provide a consistent operating model for operators to just install and run all of this stuff. Super nice. I love in 2018 we're past some of the press conversation around who's going to win or there's only going to be one kind of way of doing one stack that's going to win and Kubernetes versus whatever. And that was a conversation a few years ago. What I love about 2018 is people are in production and live and time to value very quick and very powerful and very deep and like big data apps, right? Huge, huge footprint apps as well. So can you talk a little bit about some of your customers and also in terms of the hybrid cloud? Are we seeing our people on-prem? Are you seeing a lot of multi-cloud uses? Do apps span on-prem and clouds? You know, what are some of the use cases and patterns that you see? Yeah, so I think maybe I'll start with the one I find is most interesting, which is World Caribbean, right? If I were to ask you, what is the largest computing cluster in the world by geography, you probably wouldn't say World Caribbean. So I haven't been on a cruise in a while but apparently, I remember back in the day when I was a child, when I went on a cruise you got a daily printout of today's activities and if you want to go upgrade to a meal plan or do a tour, a scuba diving, you go line up somewhere and then you register for it and if there's enough inventory, you get to do it, right? And so World Caribbean is actually trying to move all of this into a mobile app experience where based on your preferences, based on your history, based on what's available they'll push certain campaigns to get you to, John, you really got to try the scuba diving because we got access inventory and we know you have a history of wanting to do surfing, you know, excursions and so forth. So what World Caribbean's done is create an infrastructure where they're doing test and dev on campaigns and things like that on AWS. They actually do a lot of analytics on-prem in their own data center and then when the ship is out at sea, serving those mobile applications from on-prem cloud computing environment, right? All of this on Metasphere's DCOS. And what this means is that the data for interacting with passengers and the campaigns that are available, the management of the inventory, all that data when the ship is in dock flies from a data center through a satellite through Kafka into the ship. When the ship goes out to sea, all the internet connection is used for people skyping with grandma and grandpa and all that stuff. So the ship can actually, from an edge computing standpoint, provide all the resources it needs for these personalized interaction. That's a big example, Royal Caribbean. It was a very interesting use case and I know you mentioned Netflix, Verizon, I think I saw Verizon customer video on your website. When you're talking with companies of either those sizes or like, you know, Royal Caribbean that's been around for a long time versus a cloud native like Netflix, what are some of the common data center modernization concerns that you're hearing kind of consistently across company sizes and maybe even consistently across industries? Sure, sure, I think that's a great point. I think some of the early, earlier adapters like Netflix, Twitter, you know, they had their own way of building out their hyperscale infrastructures and so we worked very closely with them to address their needs. What we're starting to see as the technology becomes mainstream, there are a lot of common challenges that these mainstream enterprises are either not experienced with, aren't staffed for or just don't have the budget to blow a lot on these type of projects. And so what becomes a key concern is a lot of companies today recognize containerization is interesting, it's important, it has the potential to deliver cost savings and they recognize, you know, they have to move to kind of a DevOps model to deliver code very quickly. But then they also realize that we're starting to live in a always connected economy where you can't just sell a product and not expect to hear from the customer until they have a problem with it. You want to interact with them, you want to use that data to help improve the experience for the customer. How do you manage all this information, right? So the whole concept of data engineering, data operations and data science becomes really a key factor for many enterprises and for a lot of them they just don't have the resources to really address it. Now, there are many different companies that provide individual point solutions for those technologies, but how do you bring it all together in a multi-tenant way, right? How do you make sure if you have one team that's using one version of Spark and another team using a different version of Spark that they can actually share the infrastructure? And that's where Mesosphere's uniqueness has really come to front and center. We basically pull these data services the way VMware pulled the traditional monolithic applications. So the cost savings you saw from server consolidation we're doing from cluster consolidation and dramatically reducing costs while automating operations at the same time. I'd like to follow up on that a little bit. I think ever since the launch of DCOS a few years back big data was a differentiator for Mesosphere. And again, another term that's been through its own hype cycle, right? But it's real today. So can you maybe go a little deeper the consolidation piece, how are Mesosphere admins interacting with data scientists or even on the container side and the infrastructure side what do you have to do differently to make sure the memory footprints and the various big data platforms are able to be supported? Yeah, so I think big data 1.0, let's call it was really kind of a batch operating model, right? Wait till the data comes in at the end of the quarter make some recommendations on how the business can improve the next quarter, right? And you guys have all seen reports. I think Gartner talked about one where 80% of Hadoop projects have failed. And the reason for this is that it was hard to justify the benefit right up front. The cost and complexity of rolling out these projects was very prohibited. Now what Mesosphere brings is the ability to adopt many different types of these next generation data technologies, the Spark, Cassandra distributed database, Kafka message queue, TensorFlow, Elasticsearch. These are all technologies that have become increasingly popular, but the challenge for most enterprises is it's hard to have a whole team just dedicated to learning Kafka and another one on Spark and another one on Cassandra. What if your competitors hire them away, right? And how do you run all these different technologies that are cluster systems that require a lot of infrastructure? They're not designed to run together and pull together efficiently. So that's what Mesosphere is really bringing to these technologies, right? One is the ability to automate all these technologies. So instead of getting a whole team to figure out how to run stuff, it's literally one click installation or a single command on the DCOS command console. And then two, being able to run all these different types of data services in a highly pulled way so that you don't have different clusters that are turning into snowflakes that cannot be reused by other teams. And this has dramatic changes, this gives you dramatic changes in how people operate. Now, if you're a big data team at a major bank and somebody said, I want to do transactions on your infrastructure, you would probably say, no, stay out of my infrastructure because I want to make sure I have the resources to do analytics. And the same would be true for the people that are actually doing the real-time transaction processing with customers. What have I told you I can give you a way to do application-aware automation so that these services can be automated very easily? And two, these resources can share an infrastructure while maintaining a resource guarantees. Now all of a sudden, the individual functional leads or business unit leads would go, okay, I'm okay with sharing resources with these other BUs, especially if it gives me the benefit over time of helping different BUs cross-pollinate information. A whole different way of interacting with big data, right? And actually making it useful. Maybe forcing collaboration. Well, I wish we had more time, but we want to thank you so much for stopping by theCUBE, telling us what's new at Mesosphere. Sounds like never a dull moment. Oh, absolutely. Thank you very much. We want to thank you for watching theCUBE. I'm Lisa Martin with John Troyer from Duckercon 2018. Stick around, John and I will be right back with our last guest.