Re: TKPROF output -- high current gets for INSERT

On Thu, 30 Dec 2004 08:01:53 -0600, Cary Millsap <cary.millsap_at_hotsos.com> wrote:

> Doesn't each RE constraint use an index?
> 
> Cary Millsap

I would have thought so.

>> Bitmap indexes on FK'S for that
>> partition is made unusable before we run the insert stmt.
>> There are 13 refrential integrity constriants for this fact table that
>> are enabled during this insert operation..

Wouldn't making the indexes unusable implicitly disable the constraints? Or, to put it another way, if the constraints are enabled doesn't that mean that the indexes must be usable?

Or am I missing something?

Looking at the information in the original mail the thing that struck me (having been to lots of talks where various luminaries have made statements to the effect of "forget your hit ratios, look at your waits") is that the system seems to spend an inordinate amount of time (12 seconds) waiting for the client to talk to it. Is 12 seconds significant in the overall time the transaction takes to run?

Having to get nearly 20 blocks per row does seem a bit excessive.

Stephen

--
http://www.freelists.org/webpage/oracle-l