RE: db_file_multiblock_read_count causing full scans to take longer?

> > By flushing the buffer cache before your testing,
> > aren't you going to be testing and optimizing based
> > on an unrealistic scenario?
>
> Possibly.
>
> > Why tune for all blocks being uncached if that will
> > not be the typical case?
>
> Because this is the only way to find out what the infrastructure is
able
> to deliver.
>
> In addition:
> - If you set db_file_multiblock_read_count for best performance you
will
> eventually achieve best performance.
> - If you don't set db_file_multiblock_read_count for best performance
you
> will *never* achieve best performance.

I forgot another point... If parallel scans are used as well, then the cached blocks no more influence the performance of the I/O operations.

Regards,
Chris

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