Thread: Re: [GENERAL] big bad join problems

Re: [GENERAL] big bad join problems

From
Bruce Momjian
Date:
> On Mon, 1 Feb 1999, Bruce Momjian wrote:
> 
>     Actually, it's still fairly slow.  I've got two things that my
> view is pulling back now, and it's taking quite a while.  I'm going to
> load up a bunch of data and see if it gets any slower.  Is there an actual
> problem happening here?

With the GEQO problems people were having, I have modified the default
GEQO start table count from 8 to 6.

People are having trouble at values of > 6 for a while, but
someone(Vadim?) objected to setting it to six in the past.  With two
people having problems today, I wanted to lower it.

--  Bruce Momjian                        |  http://www.op.net/~candle maillist@candle.pha.pa.us            |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026
 


Re: [HACKERS] Re: [GENERAL] big bad join problems

From
Vadim Mikheev
Date:
Bruce Momjian wrote:
> 
> > On Mon, 1 Feb 1999, Bruce Momjian wrote:
> >
> >       Actually, it's still fairly slow.  I've got two things that my
> > view is pulling back now, and it's taking quite a while.  I'm going to
> > load up a bunch of data and see if it gets any slower.  Is there an actual
> > problem happening here?
> 
> With the GEQO problems people were having, I have modified the default
> GEQO start table count from 8 to 6.
> 
> People are having trouble at values of > 6 for a while, but
> someone(Vadim?) objected to setting it to six in the past.  With two
> people having problems today, I wanted to lower it.

Yes, it was me. I don't object against 6, but just remember that
there were other people having troubles with GEQO and this is
why table count was increased from 6 to 8.

Vadim


Re: [HACKERS] Re: [GENERAL] big bad join problems

From
Bruce Momjian
Date:
> Bruce Momjian wrote:
> > 
> > > On Mon, 1 Feb 1999, Bruce Momjian wrote:
> > >
> > >       Actually, it's still fairly slow.  I've got two things that my
> > > view is pulling back now, and it's taking quite a while.  I'm going to
> > > load up a bunch of data and see if it gets any slower.  Is there an actual
> > > problem happening here?
> > 
> > With the GEQO problems people were having, I have modified the default
> > GEQO start table count from 8 to 6.
> > 
> > People are having trouble at values of > 6 for a while, but
> > someone(Vadim?) objected to setting it to six in the past.  With two
> > people having problems today, I wanted to lower it.
> 
> Yes, it was me. I don't object against 6, but just remember that
> there were other people having troubles with GEQO and this is
> why table count was increased from 6 to 8.

Do you remember what problems?

--  Bruce Momjian                        |  http://www.op.net/~candle maillist@candle.pha.pa.us            |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026
 


Re: [HACKERS] Re: [GENERAL] big bad join problems

From
Vadim Mikheev
Date:
Bruce Momjian wrote:
> 
> > Bruce Momjian wrote:
> > >
> > > > On Mon, 1 Feb 1999, Bruce Momjian wrote:
> > > >
> > > >       Actually, it's still fairly slow.  I've got two things that my
> > > > view is pulling back now, and it's taking quite a while.  I'm going to
> > > > load up a bunch of data and see if it gets any slower.  Is there an actual
> > > > problem happening here?
> > >
> > > With the GEQO problems people were having, I have modified the default
> > > GEQO start table count from 8 to 6.
> > >
> > > People are having trouble at values of > 6 for a while, but
> > > someone(Vadim?) objected to setting it to six in the past.  With two
> > > people having problems today, I wanted to lower it.
> >
> > Yes, it was me. I don't object against 6, but just remember that
> > there were other people having troubles with GEQO and this is
> > why table count was increased from 6 to 8.
> 
> Do you remember what problems?

No. Either the same as now (long planning) or bad plans
(long execution).

Vadim


Re: [HACKERS] Re: [GENERAL] big bad join problems

From
Dustin Sallings
Date:
On Tue, 2 Feb 1999, Vadim Mikheev wrote:

# > Do you remember what problems?
# 
# No. Either the same as now (long planning) or bad plans (long
# execution). 
I don't have enough data to know whether it's planning well or
not, but I can't do two queries at the same time for lack of RAM with
GEQO, and the web browser times out without it.  :)  I'm torn.  If only I
could store a query plan. (subliminal message)

--
SA, beyond.com           My girlfriend asked me which one I like better.
pub  1024/3CAE01D5 1994/11/03 Dustin Sallings <dustin@spy.net>
|    Key fingerprint =  87 02 57 08 02 D0 DA D6  C8 0F 3E 65 51 98 D8 BE 
L_______________________ I hope the answer won't upset her. ____________



Re: [HACKERS] Re: [GENERAL] big bad join problems

From
Bruce Momjian
Date:
> > > Yes, it was me. I don't object against 6, but just remember that
> > > there were other people having troubles with GEQO and this is
> > > why table count was increased from 6 to 8.
> > 
> > Do you remember what problems?
> 
> No. Either the same as now (long planning) or bad plans
> (long execution).

My rememberance was that GEQO was slower for some 6-table joins, so it
was recommended to keep it at 8.  Tom clearly is on the proper track in
checking the number of indexes when using GEQO.  That should allow us to
set a proper value that will use GEQO in most/all cases.

--  Bruce Momjian                        |  http://www.op.net/~candle maillist@candle.pha.pa.us            |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026