Hi itgirl, I can't help first sharing that I am a trail runner and would some day love to run a marathon. Regarding your company designing clothing, I knew the day had come when I would not have to deal everything from chaffing to poison ivy protection, and there would be a company like yours that uses technology to make lives of athletes meaningful. I am also a dirtbag rock climber (never mind my saying that), and can volunteer myself to test your running gear on ridge runs. I could desperately use bacterial protection that adapts to my shower cycles. All this super motivates me to share how you can use Xcalar for your business. Now, the answer to your question -- Does Xcalar run on Hadoop/Spark -- is yes/no. Get the point? Yes for running on Hadoop and no for Spark. In fact, Xcalar hypervises Spark and gives it a real engine --a relational compute engine--to make it run 100 to 1000X faster. Spark apps run natively on Xcalar. I just tried Scala and Spark SQL and the damn thing just got shredded. Like I mean I am used to taking a bathroom break and then coffee and a walk and all that. But this xcalar engine thingy seemed to have lit a fire underneath Scala's arse. I know you don't mind my saying that because you folks are in the business of clothing that piece of human anatomy with technology fabric? Technological fabric? I can never get the Engloosh right.
Digressing...xcalar leverages Hadoop with what I believe they call True Data in Place or something technology. I think they mean using HDFS to run and not shuffling your data around or placing it for locality or making copies for joins. Deja vu, huh? Looks like we have come a whole circle from relational databases to mapreduce back to relational compute at scale. Hey, I want my shuffles back, rofl. Give them enough time to mess up and they will do a few more circles.