Re: Oracle SQL Condition Question...

In article <6hglsp$s0d$0_at_207.17.114.24>,   "Todd Marshall"
<laz777_at_semaphore.com> wrote:
>
> Your problem is the 'not in' part of the
 query. It's always much faster to
> find out if something 'is' in a table

 rather than not in a table. Without
> some background on your application, I
 can't provide you with a reasonable
> alternative though....
>
> Bhavesh

 Gosar wrote in message <6hg4p8$se0$1_at_engnews2.Eng.Sun.COM>...
> >Hi,
> >
> >I

 have 2 tables
> >
> >tab1 tab2
> >==== ====
> >.. ..
> >.. ..
> >f1 f2
>
>.. ..
> >.. ..
> >
> >Both the tables above contain approx. 60,000 Records
> >and have indexes on f1 and f2.
> >
> >I have an SQL which is the foll..
>
>
> >Insert into tab3 ( select * from tab1 where tab1.f1 not in
> > (select

 distinct tab2.f2 from tab2));
> >
> >The problem is this SQL takes around

 40-45 hours to execute...
> >
> >Am i missing something here...Or any

 Workarounds Possible...??
> >
> >TIA,
> >Bg
> >
>

Seems like I remember
seeing a tip somewhere that doing a
NOT EXISTS is quicker than a NOT IN. (I couldn't find the
material to verify.) Try rewriting the insert subquery as
follows:

select *
from tab1
where not exists
  (select 'x'
  from tab2  

where tab1.f1 = tab2.f2)

-----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/ Now offering spam-free web-based newsreading Received on Tue Apr 21 1998 - 00:00:00 CDT