Achieving optimal query performance in a distributed time-series DB on Postgres (blog.timescale.com)