Re: sql-bench

On Wed, 2006-09-13 at 14:36, Merlin Moncure wrote:
> On 9/13/06, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > IIRC, with these settings PG 8.0 seemed to be about half the speed of
> > mysql 5.0 w/myisam, which is probably somewhere in the ballpark of the
> > truth for tests of this nature, ie, single query stream of fairly simple
> > queries. If you try concurrent-update scenarios or something that
> > stresses planning ability you may arrive at different results though.
> > I have not retested with more recent versions.
>
> if postgresql uses prepared statements for such queries, it will
> roughly tie mysql/myisam in raw query output on this type of load
> which also happens to be very easy to prepare...afaik mysql gets zero
> performance benefit from preparing statements This is extremely
> trivial to test&confirm even on a shell script. [aside: will this
> still be the case if peter e's planner changes become reality?]
>
> another cheater trick benchmarkers do to disparage postgresql is to
> not run analyze intentionally. Basically all production postgresql
> systems of any size will run analyze on cron.
>
> another small aside, I caught the sqlite people actually *detuning*
> postgresql for performance by turning stats_command_string=on in
> postgresql.conf. The way it was portrayed it almost looked like
> cheating. I busted them on it (go to
> http://www.sqlite.org/cvstrac/wiki?p=SpeedComparison and look for the
> remarks right below the results)

They're running autovacuum, which requires that, doesn't it?

I'd rather them be running autovacuum than not vacuuming / analyzing at
all. And autovacuum is a pretty realistic setting for most databases (I
use it on my production machines.)