RE: Buffer Sort explanation

The temp_disable did get rid of the TEMP table
creation step in the star_transformation plan, but
tkprof still shows a very high BUFFER SORT, and the
query failed w/ 'snapshot too old' after 3 Hrs. With
the temp table step, it did finish in 2 1/2 hrs or so.

=====
      105       BITMAP MERGE
   352207        BITMAP KEY ITERATION
128553935           BUFFER SORT
  1627265          TABLE ACCESS BY GLOBAL INDEX ROWID
OBJ#(1433369) PARTITION: ROW LOCATION ROW LOCATION
  1627265           INDEX RANGE SCAN OBJ#(1433605)
(object id 1433605)
   352207         BITMAP INDEX RANGE SCAN
OBJ#(3441108) PARTITION: 1 177 (object id 3441108)


--- Christian Antognini
<Christian.Antognini@xxxxxxxxxxxx> wrote:

> Hi
> 
>  
> 
> >  Rows       Operation
> >  ========= 
> =======================================
> >a)       105 BITMAP MERGE
> >b)    351549    BITMAP KEY ITERATION
> >c) 127009880     BUFFER SORT
> >d)   1607720      TABLE ACCESS FULL
> SYS_TEMP_4254956840
> >e)    351549   BITMAP INDEX RANGE SCAN
> OBJ#(3441108)
> 
>  
> 
> I'm not able to explain you what the BUFFER SORT
> exactly does in this case (it should only get the
> data from the temporary table and prepare them for
> the "join" with the index on the fact table...
> Anyway, each time I have a problem with a star
> transformation with temporary tables I try to
> disable the temporary tables (they were/are buggy!).
> Therefore try
> star_transformation_enabled=temp_disable.
> 
>  
> 
>  
> 
> Regards,
> 
> Chris
> 
>  
> 
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 
--
http://www.freelists.org/webpage/oracle-l

Other related posts: