Just back from a few days vacation to some great news: PostgreSQL now supports Parallel Aggregation.
Parallel scans sound great, but its parallel aggregation and joins where we need parallel query most of all. So this latest step is exactly what we need to take PostgreSQL to the next level.
David Rowley’s latest version of parallel aggregation has now been committed, with results showing its effective at radical improvements for aggregation, based upon tests for TPC-H Query 1. From what I’ve seen its about x15.6 times faster on a 16-way CPU server.
(Surely, Postgres-XL must now be dead? Far from it. If you want a fully working system based upon modern open-source PostgreSQL, its still your only choice and will be for some time.)