We'll definitely be doing a DIRECT=Y, and CONSISTENT=N, because we'll
only do the export while people are reading the data, not while it is
being loaded. I'm going to do some research on my linux computer by
simulating a ton of select statements. Then for 1 hour I will monitor
it with statspack (probably every 10 minutes) and top. Then I'll do
the same and run export for 1 hours and monitor it with statspack and
top again, then compare the differences. Does this sound like a good
simulation, or should I add more to it (no DDL or inserts or deletes,
etc because that won't be going on in the real world either).
Thanks for the great responses so far,
Dustin
Received on Tue Apr 25 2006 - 12:17:55 CDT
