In article <75p1tu$vdv$1_at_nnrp1.dejanews.com>, acedba_at_my-dejanews.com
writes
>Unfortunately, partition elimination seems to occur under very limited
>circumstances ... such as when the moon is green and Jupiter aligns
>with Mars or some such nonsense. That's what I'm trying to find out --
>what criteria, or under which circumstances, is the optimizer able to
>eliminate partitions from access paths.
>
Its a bug, it should be when the moon is blue.
But seriously, the optimiser eliminates partitions
when you have eliminated then by you query.
Which is to say, if you perform a query specifying
some criteria on the partition key column(s) then
the optimiser will eliminate partitions which it
"knows" does not contain the info you want.
ie: table partitioned by col1, col1 values are A,B,C
giving you three partitions. If you search where
col1=A and <other stuff>, it will eliminate partitions
2 (B) and 3 (C).
Have a play with explain plan.
Steve
p.s
I believe 8.1 (i if you like) will have other methods
of partitioning than the quite restricted rule we
find in 8.0.*
p.p.s
There endeth the paper.
--
Please remove nospam from my address to Reply.
Received on Fri Dec 25 1998 - 17:39:50 CST
