Regular SQL is not fast enough and lacks optimal data structures to handle the data. Manufacturing execution systems, realtime monitoring of digital control systems & SCADA, monitoring the power grid, smart meters, oil & gas, petrochem, process manufacturing, all of these areas where you have intricate processes and digital controls generating a large amount of data in real time. StreamInsight lets you work with the event streams without having to necessarily worry about the underlying storage mechanism or latency. You can perform computations incrementally, in near real time, to do analysis without having to, for example, calculate an aggregate over 2 billion events. There are a number of established players who have the data storage piece down, and who "want" to be able to serve the analysis / BI needs on top of it. They have their own ways of providing for the incremental calculation. Mostly they aren't that great because there hasn't been enough competition to drive a lot of innovation. StreamInsight + the BI stack will fill a huge need in the market for which there is a lot of demand. Maybe I'm just a geek but if this performs well it is going to be a really neat piece of kit and crack open a lot of opportunities.
Microsoft Manufacturing Home
Microsoft Power and Utilities Blog