 Okay, welcome back. This is theCUBE. We're live here in Silicon Valley in San Jose Convention Center, heart of Silicon Valley. This is Hadoop Summit. We're here live for two days. This is theCUBE. theCUBE is Silicon Angle and Wikibon's exclusive coverage of Hadoop Summit. It's our flagship program. We go out to the events, extract a signal from the noise. I'm John Furrier, the founder of Silicon Angle and I'm joined by co-hosts. I'm Dave Vellante at wikibon.org. Tomer Shiran this year. He is the Vice President of Product Management at MapR. MapR is a company that has been going hard after this big data in Hadoop business for quite some time now. Adding innovations, making Hadoop Enterprise ready is really what MapR is all about. Tomer, welcome to theCUBE. Thank you. So give us the update. You guys, as I said, you kind of just don't get involved in all the urinary Olympics, I like to call it. You just kind of focus on the customers. You dive right in, trying to solve problems. Give us the update on MapR, M7, which is your platform. How are things going? Yeah, things are going great out in the field. A lot of new customers. The things we do at MapR is we make Hadoop Enterprise grade. We bring customers all the innovation that happens in the open source community. We combine that with our own innovation to make the platform more Enterprise grade and we provide significant architectural advantages if you look at things like being POSIX compliant and offering full, random read-write access so customers can just mount the cluster that they would mount a giant NAS and integrate all their standard tools. Yeah, there's a lot of things there for the customers. We're way up front on that one, right? Yeah, I mean, it's all about the underlying architecture that enables you to do those things. It's not something you can add kind of as an incremental patch to the existing systems. Maybe add some color to that, if you would. Talk about the underlying architecture. I'm inferring from what you said that there's an architectural flexibility. It allows you to add features in a way that are not bolt-ons. I'm inferring a lot from that statement, but I wonder if you could just double-click on that a little bit and share with our audience what you mean by that. Sure, we've, talking to many, many companies that use Hadoop in a lot of the largest enterprises now use MapR as a distribution. What we've found is there are a lot of limitations with kind of the other Hadoop distributions in terms of, for example, only being able to append the files, essentially, HDFS being a read-only file system. And so MapR has really come up and solved those problems by providing the ability to do random reads and writes and the ability to expose a standard storage interface. So it's very easy to use the platform. And those kinds of innovations, like having an NFS interface, are only possible when you have the underlying random read-write capability that we've built. And that's not something that can be just bolted onto HDFS. You know, the same things apply for all the enterprise-grade business continuity. If you look at things like high availability, not having a name node or snapshots for a point-in-time recovery with full consistency or disaster recovery across data centers, those are all features that are unique to MapR and they're enabled by the advanced architecture. Tom, I've got to ask you, obviously, MapR M7, big news here. Also, the joint testing with Fusion IO company, we've been covering since they've been private, now public pioneering the Flash SSD space for, you know, on server side, among other things, just really changing the game on IO and then these, you know, we call up, you know, software-led infrastructure, software-defined infrastructure. What have you guys found? Because obviously, writes and reads is a big debate. You know, get these data lakes out there being talked about. HBase makes a great opportunity to take advantage of those data lakes. What did you guys do in M7 and with Fusion? What have you guys, can you share with the folks some color behind the Fusion relationship and that announcement? Yeah, absolutely. So, you know, MapR's M7 edition is all about bringing together the different types of the Hadoop workloads as well as the NoSQL workloads with the ability to run HBase applications in a production environment. So M7 is all about enterprise-grade HBase in addition to Hadoop and having that one platform. From a performance standpoint, you know, we've solved all the inefficiencies that otherwise exist in the software layer and with M7, it's really all about driving, you know, we're really able to drive the hardware at its raw speed. So if you look at technologies like SSDs and specifically Fusion I.O., that brings to the table, you know, additional performance improvements at the hardware layer that MapR can uniquely take advantage of. If you look at other Hadoop distributions, they're bottlenecked in the software and I've seen blog posts talking about how Hadoop can't benefit from SSD. I think we heard that today it won't benefit from SSDs in the next three years and really that's because, you know, there's inefficiencies at the software layer that needed to be addressed and MapR has addressed those. And I think the other point there is, you know, most customers, they don't want to run the entire cluster just with SSDs and so MapR gives them the ability to provide tiered storage where they can do some workloads, for example, the HBase type workloads on SSD and other workloads on the spinning disks. It's funny, we've been, you know, it's funny for watch and interesting to watch, not so much funny, but it's just funny to watch that the early criticisms of MapR and that when you guys were founded and launched, oh, well, they're non-standard, et cetera, et cetera. You know, you're smiling, you know what I'm talking about. But now everyone's catching up to you guys with this enterprise grade messaging. So I really want to ask you, you guys have stayed true to your mission on the enterprise side, really building, using open source, building a product for the enterprise. So just comment anecdotally if you can, just on everyone else kind of catching up, because Merve said it's the bike race where everyone kind of catches up to someone else. Or have they caught up? Are you extending your lead? And what have you found in the enterprise that's going to keep you ahead of the pack in terms of differentiating on that enterprise grade? Yeah, so certainly the MapR... Take your choice. There's a lot of questions there, but the MapR message around being enterprise grade and the features and the capabilities we've delivered to accomplish that. You know, we see our competitors talking about wanting to be enterprise grade or providing features that may sound similar to MapR. So to give you an example, some of our competitors talk about H-based snapshots. Well, snapshots are all about point in time recovery. Well, H-based snapshots don't have any consistency in them. So they're not snapshots. And if you read the JIRAs, they talk about maybe we should rename these fuzzy snapshots so that we don't confuse the customers and why don't we just let's leave the name and we'll document it and things like that. So I think it's great validation for our approach to the market, for our strategy that everyone talks about enterprise grade and our competitors want to have their features called similar to MapR's features. But there's no silver bullet though in the enterprise. I mean, it's clearly, it's not a one size fits all marketplace. I mean, do you agree with that statement? Do you see it differently? I mean, there's no one Hadoop. It's a lot of different solutions depending upon the use cases, right? Or how do you see that? Well, I think Hadoop in itself and specifically a lot of the things we've done are about enabling more and more use cases to run on the platform. You know, one of the things that sometimes people don't realize is that with MapR you get the benefit, all the advantages of the open source and the open source community and any new open source projects that happen or that come up or are integrated into the MapR platform and our customers enjoy our 24 seven support on those. And then we combine that with our own innovation so the customer gets the benefit of both worlds. So talk about that a little bit. So let's talk about that in the context of yarn. So how does that statement apply to something like yarn? Take us through sort of a practical rollout from a product standpoint. Yeah, so you know, there's, you know, I think one of the presentations this morning talked about Hadoop being not one project but many projects. And so yarn being one of the projects in the Hadoop ecosystem. We at MapR, we're really excited about yarn. We'll have a GA release with yarn later this year. Yarn is all about expanding and the use cases for Hadoop and enabling interactive queries as well as batch and streaming and so forth on the single platform. And by bringing that innovation that's happening in the open source community to our customers, to MapR users, they can take advantage of that. For example, projects like Apache Drill which are about interactive SQL queries, those are baked into yarn. Those run within the yarn context and they're part of that. And I think what you'll see going forward is that yarn will enable new types of workloads running within Hadoop and by having that unique underlying platform that MapR has which allows any application to access the data, not just applications that were designed for Hadoop, all of a sudden you'll see things that can run in yarn like MPI which will only be able to access the data when they're running on a MapR distribution. Okay, so classic example of you guys adding value to an open source component. All right, so what's next for you guys? What's the roadmap look like? Give us some hints. Yeah, I mean, if you look at the MapR, we traditionally haven't talked that much about our futures. We haven't set up conferences and talked about futures of Hadoop. We talked about futures of Hadoop. Futures of Hadoop. What kind of gaps are you guys going to fill? Right, our vision for Hadoop is the one platform for big data. As these clusters grow bigger and bigger and MapR now has customers that run over 1,000 servers in a cluster, that's where the data lives. And it's not really possible anymore to kind of move the data around to different other systems for different types of access to that data. And so it's all about expanding the use cases and the types of access you can have to that platform and then continuing to provide more and more enterprise-grade capabilities so that customers, our enterprises, are comfortable using this in the most mission-critical environments. Tellmer, thanks for coming on theCUBE. Vice President of Product Management at MapR, obviously one of the big pioneers here we call them, we're part of the big three, Hadoop, CloudEra, Hortonworks, MapR. You guys have been there from the beginning. Congratulations, great company. We've been following your progress, big fan. Again, we think this model's going to evolve faster and faster and faster. This is theCUBE, this is Hadoop Summit. I'm John Furrier with Dave Vellante. We'll be right back with our next guest after this short break.