 Hi everybody, we're back. This is Dave Vellante of Wikibon. This is theCUBE, SiliconANGLE's continuous coverage of Strata, O'Reilly Media's flagship program around big data. Now several months ago, we introduced you from Wikibon headquarters to a company called WinDisco, which many of you had not heard of. We started writing about them a little bit. They really caught our attention. And we had some folks from WinDisco on yesterday talking about what they're doing, very specifically solving some of the harder infrastructure problems around Hadoop. In particular, we often hear Hadoop is not enterprise ready, everybody's working on that problem. Well, WinDisco is a company that has come out with a very unique distribution for Hadoop, and also some unique IP that allows for active, active infrastructure really to support mission-critical applications. So we're back here with David Richards, who's the CEO, and Jugane Sundar, who's the CTO and VP of Engineering for the big data side of the house. Gentlemen, welcome back. Thanks, Dave. Thank you. We talked yesterday about your Hadoop distro, the reaction that you've had in the market in place, why you're doing it, and we talked specifically about the problem that you're solving there, in particular around supporting mission-critical production systems in Hadoop. So I want to start, David, with you and just find out what the reaction has been to that at the event. There's a lot of noise going on around distros, some other big vendors. What are people telling you when they come by your booth and talk to you in the hallways? So first of all, I've put that into two camps. I've put the other vendors at the trade show and then potential customers' enterprises. So taking the first one first, I think there was a lot of skepticism prior to us coming here today because we hadn't announced any products. We sort of said, this is what we're going to do, this is what we can potentially do. And actually, the demo, which you guys came and did a film of this afternoon, this morning, sorry, has just gone down so well. I think it's a phenomenal demo, it's live, we've got our partners from Hive and Fusion, I understand as well, and we've been demonstrating, taking a server offline, replicating a server, crashing a flood, an earthquake or whatever, and then how Hadoop recovers very quickly out the one disco active Hadoop recovers without any downtime. I think the reaction from other vendors was like, wow, these guys, this is the real deal, they can actually do this. Now, from customers, we're getting the reaction of, well, first of all, we're hearing that there's not a great deal of substance in terms of new products. There's a lot of new partnerships, I think, been announced, but in terms of new products, I don't think there's a hell of a lot that have actually been announced at the show. I think that puts us into a good place with enterprises that are kind of coming around and saying, you know, we've been trying Hadoop for the past 12 months, maybe in some sort of Skunkworks project, maybe using EC2, maybe using Lambson's Lastic Cloud, but now we want to go into production, how are we going to do this? And we've, I think we saw, maybe, I think we've seen 150 different enterprises come by our booth that we've pre-qualified in so far, so I think we're very satisfied with the show as a whole, actually. But David, we talked yesterday about that five to one, ratio five to one, sort of tire kicking, experimentation to one deployment. We talked about, you know, your objectives of moving that needle. Do you feel like after talking to those 150 that your objective can be realized in the near to midterm? Is that something that you, you know, gathered some more data around, and I wonder if you could share that with us? Yeah, I think so, I think what we've seen is that they're missing two or three critical pieces to the puzzle. One of those pieces I'm delighted to say is high availability, I'm bound to say that, but it is, it's high availability, guys. What's the enterprise? Yeah, exactly. You know, I mean, I was having a conversation with one of the cloud vendors who said, yeah, we have a few outages, and that's like saying, well, my hot pacemaker goes out for 20 minutes and that's okay, right? No, it's not. I lose the kids every now and then. Yeah, eventually find out. I like that, I know what you mean. And secondary security, you know, and I think those are the two or three things that I think they're looking for that they've, you know, 2012 I think was the year of the prototype, and 2013 I think it was the year when they're going to try and push these things into production, and we're hearing that, not just anecdotally, but customer after customer after customer is coming back and telling us that. Yeah, and security has been a big theme. We saw Intel announce its own distro, and I think there's really a big security angle there, which is good, I think that's good for the industry, especially when you start talking about the internet of things. Jagane, I wonder if you could talk a little bit about the vision from a roadmap standpoint. Share with us what you see coming. So from our company's perspective, we see our active, active replication as having unique value in this marketplace. So we've started by demonstrating the capability in the name node. We can take down a name node, and the jobs continue to run. We can guarantee HPACE availability, even if a log role is happening while the name node goes down. The next step, a logical step for us is to extend that capability to other single points of failure in the Hadoop ecosystem. We're going to start with the HPACE master. We're going to implement our high availability in the Yarn resource manager. Once that is done, we feel that most of the critical components of Hadoop will remain up no matter what. After that, we have some file system enhancements that are really interesting. We intend to bring our cross data center active, active replication into HDFS, and we'll be building what we are calling the world file system, where you can have a single Hadoop that spans multiple data centers. Those are the things in our immediate roadmap. So you're talking about longer physical distances. What we typically associate with asynchronous distances, is that right? Absolutely. We expect to implement this. Remember, our company's core intellectual property is an implementation, a patented implementation of PAXOS that's capable of running over long distances across continents. This is not a New Jersey, New York data center type of replication, and we will apply that to Hadoop. When I think about active examples today in the marketplace, I mean, one is, for instance, EMC's Vplex, and that's really a local synchronous distance capability. They've announced some stuff for distance, but it's very unclear if and when that'll ever see the marketplace. So you're talking about a really hard problem. Actually, I met with somebody from Berkeley last night, and they said, we don't have enough hard problems to solve, and I said, what about the speed of light problem? And they said, oh, that's easy. We can solve for that algorithm. It went way over my head, but essentially have you figured out a way to solve for that speed of light problem? Absolutely. I mean, the USPTO did a great deal of investigation before they gave us the patent, and the patent is for that specific enhancement. We can do this over the van. We can bring Hadoop namespace problems across the van. Typical installation today would have two separate instances of Hadoop and two data centers, and they run something like this CP to copy the data over. The little known secret is that the data wanders over time. So you may have a thousand files that are slightly different, 10,000 files that are different six months from now. There's a problem now. Which one is the correct file, and who gets to decide what needs to be deleted? That's a problem that's very labor intensive. We have a single Hadoop across the data center, which means there is no question of two different views of what the files are, and that is unique value that enhances our value proposition. Okay, and how much can you tell us about the secret sauce? Are you solving that algorithmically? Are you using some kind of a racer coding methodology? All of the above, none of the above? So it's basically an implementation of PAXOS that is algorithmically capable of running across wide area networks, and that is the core essence of our intellectual property. A racial coding is interesting, but it's a solved problem. It's one that we don't have a patent on, and it's not necessarily the right application for this particular problem. Okay, so David, what will this capability mean for customers, and to take it back to a business value standpoint? So imagine a world where you're not reliant on a single point of failure, which is the data center itself. Imagine a world where you have a data center in the United States, a data center in Europe, a data center in India, a data center in China, in fact a data center in the Antarctic, if you wanted to put one up there. We don't have any physical limitations, which means that in the event of those nasty earthquakes, floods, natural disasters that affect an entire region, you're not subjected to outage because of a regional problem, like a power outage or a flood or an earthquake. So I mean, what's that worth to a business? Well, what's any outage cost to business? There's a huge opportunity cost. Well, but yeah, now I want to pick at that a little bit. So it's a really global infrastructure capability at scale that you're providing, where the building is not a single point of failure. But you can sort of engineer that today by spending a lot of money, but you can't place all your applications on it. And the time to recovery is an elongated time to recovery. Our time to recovery is near as damn it, zero. So you're talking about much better economics underlying this whole thing with a superb RTO. And of course, RPO of as close to zero as you can possibly get. I guess there's no such thing as zero RTO. That's the speed of life problem. Yeah, right, right, right, right. We've got the Berkeley guys to do it. That's great. Well, I really have been impressed with, as I said yesterday, your knowledge, your insights, your ability to translate that into value. So appreciate you guys doing what you're doing for the industry, coming on theCUBE, supporting SiliconANGLE, Wikibon. So thanks very much again for taking the time out. Good luck with the rest of the event. We'll see you around the show, I'm sure, and around the industry. Perfect. Thanks, Dave. Good to see you again, David. Good to see you again. Thank you. Thank you. All right, everybody, keep it right there and we'll be back with our next guest. We're live. This is theCUBE from Strata. Looked at all the programs out there and identified a gap in tech news coverage. There are plenty of tech shows that provide new gadgets and talk about the latest in gaming, but those shows are just the tip of the iceberg and we're here for the deep dive. There's a difference between technology consumers and those who live the business day today and our viewers recognize that. The market begged for our program to fill that void. We're not just touting off headlines. Our goal is to provide you with a story but we also wanna analyze the big picture and ask the questions that no one else is asking. Our guests aren't just here to provide commentary. We work with analysts who know the industry from the inside out. The tech business isn't new but many networks treat it as if it is and really barely scratch the surface on technology coverage. We follow the expansion of the cloud and the evolution of big data. We're covering new enterprise from startup to IPO and every move in between. So what do you think was...