Re: [HACKERS] Postgres Speed or lack thereof
От | John Holland |
---|---|
Тема | Re: [HACKERS] Postgres Speed or lack thereof |
Дата | |
Msg-id | Pine.GSO.4.02.9901181026200.19682-100000@europa.isr.umd.edu обсуждение исходный текст |
Ответ на | Re: [HACKERS] Postgres Speed or lack thereof (Vadim Mikheev <vadim@krs.ru>) |
Список | pgsql-hackers |
I tried to do this..... What I've done is create a plpgpsql function that does the inserts. There is a begin/end in the function of course...I also tried doing it with a begin work before calling the function in psql, and it did seem that this worked as it should - ie a rollback work would get rid of the 40000 inserts. Still, it took about 3.5 minutes, and Oracle does it in about 30 seconds. factor of 7..... On the plus side for Postgres, trying to find some queries against the data for other speed tests, I found that it seemed to respond quickly - not much thinking to do with one table of integers I guess. Actually I couldn't come up with a query that would take very long - need another table I suppose. I found that UPDATE TEST SET A = B -1 took only 20 seconds for 40000 records. The real plus is that Oracle couldn't do that transaction because it was running out of rollback segment space - something I have fixed in the past but I recall being a pain to fix - need to make bigger rollback datafiles or whatever. Any suggested speed tests would be appreciated. > > Try to use BEGIN/END to run all inserts in single transaction > and please let us know results. > > Vadim >
В списке pgsql-hackers по дате отправления: