I tried inserting 15000 tuples each in two tables via pgsql and timed it.
After insertion I also created btree indeces on both tables.
Some results:
First try was with a 6.4.1 version without -F option and all inserts in one
transaction:
real 2m11.084s
user 0m1.760s
sys 0m1.680s
Then I switched to 6.5 without -F:
real 2m35.833s
user 0m2.900s
sys 0m3.910s
So some of the latest changes cost quite a lot of time. On the other hand I
was surprised to see that the size of the table file decreased by about 10%
when switching to 6.5.
Then I switched on -F:
real 1m11.573s
user 0m3.150s
sys 0m3.760s
I've seen even bigger advantages of -F.
For comparison I tried without BEGIN/END:
real 1m12.463s
user 0m3.030s
sys 0m4.480s
So, yes it is slower than just one transaction, but not very much.
Finally I tried the same script (except for changing int4 to number(4)) on
Oracle 8.0.5:
real 1m29.248s
user 0m24.460s
sys 0m4.880s
I got similar results when I tried the last time. On stuff like insertions
we are quite a lot faster than Oracle if we use -F.
All test were run on my AMD K6/350 with 64MB ram running Linux-2.2.0-pre7.
michael
--
Michael Meskes | Go SF 49ers!
Th.-Heuss-Str. 61, D-41812 Erkelenz | Go Rhein Fire!
Tel.: (+49) 2431/72651 | Use Debian GNU/Linux!
Email: Michael.Meskes@gmx.net | Use PostgreSQL!
