Can I ask why you decided to partition Postgres like this? It looks like you have time-series data based on the examples, and you wanted quick dropping, am I right?