[askdba] Re: SQL -HELP
- From: Abraham Kurian <kurian24@xxxxxxxxx>
- To: askdba@xxxxxxxxxxxxx
- Date: Thu, 9 Dec 2004 09:28:51 +0530
hi
Its possible that , the optimizer would choose the best possible path
to fetch the rows . in your case the PLAN changes from a Unique Scan
to a Index FFS ..
Can you send me the complete Explain Plan with ALL_ROWS.
send me the O/p of
select * from table(dbms_xplan.display());
Also, the details of indexed columns
regards
Kurian
On Tue, 7 Dec 2004 22:48:48 -0800, VuCanDo <vuquyen@xxxxxxxxx> wrote:
> Even I simplified the SQL:
>
> select count(1)
> from spy_score m, spy_claim_case c
> where m.claim_id = getClaim(c.claim_id) and m.claim_line =
> getLine(c.claim_id) AND c.status = 'PEND' AND case_queue = '1'
>
> From SCHEMA#1
>
> FIRST_ROWS SELECT STATEMENT Cost = 12
> 1.1 SORT AGGREGATE
> 2.1 NESTED LOOPS
> 3.1 TABLE ACCESS FULL SPY_CLAIM_CASE
> 3.2 INDEX UNIQUE SCAN SPY_SCORE_K1
>
> From SCHEMA#2:
>
> FIRST_ROWS SELECT STATEMENT Cost = 957
> 1.1 SORT AGGREGATE
> 2.1 HASH JOIN
> 3.1 TABLE ACCESS FULL SPY_CLAIM_CASE
> 3.2 INDEX FAST FULL SCAN SPY_SCORE_K1
>
>
>
>
> On Tue, 7 Dec 2004 21:17:24 -0800, VuCanDo <vuquyen@xxxxxxxxx> wrote:
> > Yes and the amount of data for SCHEMA#1 is couple millions rows bigger
> > than SCHEMA#2.
> >
> >
> >
> >
> > On Wed, 8 Dec 2004 09:35:03 +0530, Abraham Kurian <kurian24@xxxxxxxxx>
> > wrote:
> > > Is Stats are updated ? what about the amount of data each table in
> > > holding ?
> > >
> > > Regards
> > > Kurian
> > >
> > >
> > >
> > >
> > > On Tue, 7 Dec 2004 15:49:15 -0800, VuCanDo <vuquyen@xxxxxxxxx> wrote:
> > > > Hi Guru,
> > > >
> > > > I have a SQL statement and run against 2 schemas exactly the same,
> > > > but both give me a different explain plan run on same instance 9.2.0.4
> > > > on NT. Schema#2 seem to take longer then schema#1 as you can see.
> > > > What's wrong with this SQL? See below:
> > > >
> > > > select m.score_dt, count(1), sum(paid_amt), sum(allow_amt),
> > > > min(overall_score), max(overall_score)
> > > > from spy_mart_score m, spy_claim_case c
> > > > where m.claim_id = getClaim(c.claim_id) and m.claim_line =
> > > > getLine(c.claim_id) AND c.status = 'PEND' AND case_queue = '1'
> > > > group by m.score_dt
> > > > order by m.score_dt desc;
> > > >
> > > > Explain plan from SCHEMA#1 is
> > > >
> > > > FIRST_ROWS SELECT STATEMENT Cost = 22
> > > >
> > > > 1.1 SORT GROUP BY
> > > > 2.1 TABLE ACCESS BY INDEX ROWID SPY_SCORE
> > > > 3.1 NESTED LOOPS
> > > > 4.1 NESTED LOOPS
> > > > 5.1 TABLE ACCESS FULL SPY_CLAIM_CASE
> > > > 5.2 TABLE ACCESS BY INDEX ROWID SPY_MART
> > > > 6.1 INDEX UNIQUE SCAN SPY_MART_K1
> > > > 4.2 INDEX RANGE SCAN SPY_SCORE_K1
> > > >
> > > > Explain plan from SCHEMA#2 is
> > > >
> > > > FIRST_ROWS SELECT STATEMENT Cost = 27169
> > > >
> > > > 1.1 SORT GROUP BY
> > > > 2.1 NESTED LOOPS OUTER
> > > > 3.1 HASH JOIN
> > > > 4.1 TABLE ACCESS FULL SPY_CLAIM_CASE
> > > > 4.2 TABLE ACCESS FULL SPY_MART
> > > > 3.2 TABLE ACCESS BY INDEX ROWID SPY_SCORE
> > > > 4.1 INDEX UNIQUE SCAN SPY_SCORE_K1
> > > >
> > > > Thanks,
> > > >
> > > > Vu
> > > >
> > > >
> > >
> > >
> >
>
>
Other related posts:
