Dear List,
I gave been investigating how postgresql does sorting, using the query from my
last posting :
select d0.d0f1, count(f.f1) from dim0 d0, fact0 f where d0.d0key = f.d0key and d0.d0f1 between
'1996-05-01'and '1996-05-31' group by d0.d0f1 ;
There is about 5-7M of data to sort - all with the same sort key ( d0f1 ).
I noticed that 19000K of temporary sort tables were written during query
execution, so I tried using backend options ( -o ) of : -S 25000.
Whilst there was no sort tables written with this option, the elapsed time for
the query was longer ! ( 35s instead of 14s )
The platform is redhat 6.0 and pgsql 6.5.3
There is 160M ram with about 60M free during the query exec
( note that hpux 11 with pgsql 6.5.3 gives similar results too...)
Is there something I need to set or unset, or is this expected ?
Cheers
Mark
