I'm curious how much space you could save using something like pg_column_byte_packer in practice.


· · Web · 3 · 4 · 5

I'm tempted to do one last experiment on the faulty server I intend to cancel, by re-ordering the columns on Mastodon's schema and then using logical replication, to see the difference in disk space used.

Alright, I did a very cumbersome experiment on the accounts table, because it's got a lot of columns of different types and not so many rows that it would take long to copy data.

Original size: 2498 MB. Size when columns are re-ordered: 2443 MB. So that's like 2.2% savings (55MB).

Of course, every table is different. Maybe you could squeeze more out of the statuses table, which is so much larger. But I'm not sure it's worth the effort...

Sign in to participate in the conversation

Server run by the main developers of the project 🐘 It is not focused on any particular niche interest - everyone is welcome as long as you follow our code of conduct!