Parallel option slows performance - why is this?
Hi, I have a mapping that essentially does the following:
a) full table scan of some incoming table
b) inner joined with full table scan of a lookup table
c) resulting data merged into a target table based on a PK set of fields
I'm running in test - truthfully don't know the types of boxes, etc. What I see is that the mapping runs in about 6 minutes 10 seconds in parallel, but only 5 minutes 30 seconds serial. Can anyone help me understand why this is? I'd think with all the full table scans, etc. that parallel would be MUCH faster. Would this be caused by IO bottlenecks, insufficient memory, CPU bottlenecks, etc.?
p.s. the incoming dataset has about 1.6 million rows. The natural key used to join in step B consists of about 10 varchar2 columns.
Thanks in advance for the ideas! I'll go talk to the DBAs and hardware folk...but want to have some intelligent ideas going in.
Thx,
Scott