Re: speed of querry? - Mailing list pgsql-performance

Well so far I have 1.5 hours with commandpromt.com and 8 + hours with Dell
and have not seen any changes in the speed of my query.

I did move the data base to the raid 5 drives and did see a 1 second
improvement from 13 secs to 12 secs (keep in mind it runs in 6 on the
optiplex).

The dell guy ran Bonie and found 40meg per sec read/write speed for the
arrays.

He also installed version 8.0.2 (went fine on AS4 he had to uninstall 8.0.1
first).

He is going to get a 6650 in his test lab to see what he can fugure out.
I will say both commandprompt.com and Dell have been very professional and I
am impressed at the level of support available for Redhat from Dell and
postgres. As always I still feel this list has been my most useful asset,
but I am glad there are folks to call on. I am trying to go live soon and
need to get this resolved.

I told the guy from Dell it makes no sense that a windows 2.4 single proc
with 750 meg of ram can go faster then a 4 proc (3.ghz) 8 gig machine.
Both databases were restored from the same file. Same view etc.

Config files are set the same except for amount of cached ram, although
Commandprompt.com had me adjust a few items that should help going into
production, put planning stuff is basicly the same.

This view returns in 3 secs on MSSQL server on the optiplex (750 meg 2.4
box); and 6 secs using postgres on windows and 12-13 secs on the 4 processor
box. Needless to say I am very frustrated. Maybe Dell will turn up something
testing in their lab. It took a bit of perseverance to get to the right guy
at Dell (the first guy actually told me to load it all on a like machine and
if it was very much slower on my original they would pursue it otherwise it
was not an issue. I was like the machine cost 30K you going to send me one
to test that. But seriously I am open to trying anything (loading AS3, using
postgres 7.4)? The fellow at Dell does not think it is a hardware problem,
so if it is Linux (could very well be, but he seemed very sharp and did not
come up with anything yet) or postgres config (again Josh at
commandprompt.com was very sharp) then what do I do now to isolate the
issue? At least they are loading one in the lab (in theory, I cant send them
my database, so who knows what they will test). Dell changed the file system
to ext2 is that going to bite me in the butt? It did not seem to change the
speed of my explain analyze.

Joel Fradkin


Dawid Kuroczko <qnex42@gmail.com> writes:
> Basically it tells postgres how many values should it keep for
> statistics per column.  The config default_statistics_target
> is the default (= used when creating table) and ALTER... is
> a way to change it later.

Not quite.  default_statistics_target is the value used by ANALYZE for
any column that hasn't had an explicit ALTER SET STATISTICS done on it.
So you can change default_statistics_target and that will affect
existing tables.

(It used to work the way you are saying, but that was a few releases
back...)

            regards, tom lane


---------------------------(end of broadcast)---------------------------
TIP 3: if posting/reading through Usenet, please send an appropriate
      subscribe-nomail command to majordomo@postgresql.org so that your
      message can get through to the mailing list cleanly