> As long as your archiver processes are able to keep up with the log file switches, it doesn't matter what type of filesystem they're on.
I disagree. One should consider the memory load of spooling logs to a buffered filesytem. I'd recommend spooling them to a direct IO locale so you don't load down memory with a bunch of filesystem blocks of archived logs...just a waste of both processor cycles (lots of kernel sice memcpy) and memory capacity.
If this is RAC, you better put the archived logs someplace that other nodes can see them...and while you're at it, make sure you acatually *need* RAC...
but, oh, wait, we don't need filesystems anymore. I forgot -- http://www.freelists.org/webpage/oracle-l
