 Hi everyone, this is Jeff Kelly of Wikibon. I'm here on the floor of Stratoconference at Santa Clara, California. We're here with DDN and Uday Mohan who's going to talk to us a little bit about DDN, what they're doing in the Hadoop world, a recent announcement around Hscaler. Welcome to theCUBE. Thank you, thank you. Hi folks, my name is Uday Mohan. I am the product manager for Hscaler. If you know DDN, DDN is a provider of massively scalable and high-performance storage systems. So customers come to us when they have challenges at the very high scale of performance or scalability in storage systems. If you take a look at the top 100 supercomputers in the world, more than 70% run on DDN storage. So we are really known in HPC and we are basically leveraging our leadership in HPC and we are applying it to analytics. So we are really excited to announce Hscaler. This is a completely unified end-to-end Hadoop appliance that makes Hadoop really simple to manage or install. It's a complete appliance end-to-end that you can basically roll into your data center, plug in power, plug in networking, point to a data source and start crunching numbers. It's got storage, it's got compute, as well as networking all combined with the Hortonworks Apache distribution as well as a very flexible ETL engine. The Hscaler is built on the SFA12K40. This is the world's fastest big data platform, can go to 40 gigabytes per second. So this really helps us excel at Hadoop. Layered on top is a very highly optimized InfiniBand-based network. Really helps us accelerate Hadoop. So if you take a look at Hadoop, there's the map phase, there's the shuffle phase and there's the reduce phase. And the shuffle phase is the phase where there is a lot of data transfer in the backend network. And so we really accelerate that phase with the InfiniBand fabric here. And then there are highly optimized, very dense compute nodes. These are running Hortonworks distribution of Hadoop as well as a very flexible ETL engine that can connect to over 200 data sources to be able to plug in and import data really quickly. So the net result is you can manage this entire stack from a single point using DDN's DirectMon. Very flexible and intuitive cluster management system. So you can manage storage, networking as well as compute as well as the ETL engine from that single pane of glass. So the end result is we've really tried to simplify the deployment of Hadoop. We've layered multiple optimizations across the entire stack. At the storage layer, we have the SFA12K that is really fast, but we've offloaded all the data protection features from a typical Hadoop node onto storage. So if you take a look at a typical Hadoop cluster, a large part of the CPU cycles are being used to do data protection, data management. If there's a node failure, the backend network is being used to move data across. Well, we've offloaded all of that to the SFA12K and this is designed to do that really well. So we, and in the SQL architecture, can release those CPU cycles back to compute. Moving up the stack I talked about InfiniBand RDMA, it's end to an RDMA base. So we really accelerate the shuffle phase. And also if you take a look at any typical Hadoop job, about 30% of the data is not local to the node running the actual compute. So we really accelerate that by giving all nodes equidistant access to data using the centralized storage. And then there's highly dense compute. So the net result is we can really simplify deployment of Hadoop, but also lower your total cost ownership. The SFA12K can pack up to 84 drives in just for you. This is about four times denser than anything out there. Typically people use Dell and HP and they have 12 to 24 drives into you. So the net result is customers come to us and say, well, I need a 10,000 node Hadoop cluster. I may do it in say a hundred racks. With Edge Scaler, you can actually cut your data center cost by one fourth. And plus the fact that because we've built optimizations across the entire stack, we can really accelerate Hadoop. So you can do this in a lot less hardware. So we really save on your acquisition cost as well as your operational cost. All right, well, fantastic. Thanks so much for taking some time today. Really appreciate it. DDN, getting into the Hadoop market with a really interesting release with the Scaler. Thanks for watching.