Hi !
I would like to thank you all for your detailed answers and explanations.
I would give "partitioning" a try, by creating a dedicated new partition
table, and insert a (big enough) extract of the source data in it.
You are right, the best would be to try in real life !
Best wishes
Kimaidou
>
> Yours will be different, as I cannot exactly duplicate your schema or data
> distribution, but give "SELECT 1" a try. This was on Postgres 16, FWIW,
> with a default_statistics_target of 100.
>
Select 1 produces a sequential scan, like Select * did before Vacuum Full.
But if I force an index
