I did a year long project to build a flexible engine for materialized views onto 1-10TB live event datasets, and our architecture was roughly converging toward

I did a year long project to build a flexible engine for materialized views onto 1-10TB live event datasets, and our architecture was roughly converging toward this idea of "ship the code to where the indexes are" before we moved onto a different project I'm very compelled by Rama, but unfortunately won't adopt it due to JVM for totally irrational reasons (just don't like Java/JVM). Would love to see this architecture ported!