Oracle Database Discussions

Hi,

I read the amazing articles from Ixora on async IO and directIO under Solaris
and greatly improved performance following their good advise.

I've got Solaris 9, Oracle 10gR1, cooked file systems and filesystemio_options
set to 'setall' to use threaded AIO and directIO where beneficial. This is all
good but my ADDM reports now constantly show significant database time
on 'i/o slave wait' and this is always my biggest area to reduce database time.

Perhaps this is to be expected, I am using threaded AIO after all, but I find it
annoying that this known background wait event is now skewing my ADDM reports.

Have I missed something? Is this the expected behavior? Can I tell Oracle to
just get over it, live with it and stop telling me about it? I've also set dbwr_io_slaves
to the max setting without much improvement. Is this the shove that will eventually
force me to use ASM and kernel based AIO?

Darren