Kaarel Moppel puts on the lab coat and safety glasses:
In my last post I did a quick check on the performance of the newer (lz4, zstd)
pg_dump
compression options, which included setting up a small framework to download some openly available “real life”-ish sample datasets. And the general result was that, indeed – the new algos in lower levels provide the best value, especially zstd.But
pg_dump
is about compressing essentially text based data…but how about binary Postgres data? Thus the tool to test here additionally ispg_basebackup
, with its newer (v15+) compression options. So let’s see if something stands out consistently again.
Click through for the test results.