Four fails and a win at a big data stack for realtime analyticsBuilding a user-friendly app to analyze big data in real time (that is, keeping response times below 60 seconds) is a challenge. In the big data world, you’re either doing batch analytics where nobody really cares about query time (most businesses); or you’re doing streaming (Uber, Facebook and kin) where query time is critical, but data is only big on aggregate — each user only sees or uses a tin
