> Yes, I use the same approach, but I'm not aware of any such guideline
> related to fillfactor with indexes. Anyway those guidelines need to be
> written by someone, so you have a great opportunity ;-)
I did a quick test using your example. As in your test, "increasing"
values don't get any gain from a different fillfactor.
I tried a random index:
create table test_fill (id int);
create index test_fill_idx on test_fill(id) with (fillfactor=100);
insert into test_fill select (random()*100000)::integer from
generate_series(1,10000000) i;
time: 373936.724
drop table test_fill;
create table test_fill (id int);
create index test_fill_idx on test_fill(id) with (fillfactor=50);
insert into test_fill select (random()*100000)::integer from
generate_series(1,10000000) i;
time: 393209.911
not much difference...
Now I'm getting confused... is which cases fillfactor makes a difference???