Several people have made some good suggestions, and I
would advise you look at the white papers on the Netapp site as well. One thing
that will improve things for Solaris in particular is to specify
DISK_ASYNC_IO=false in the Oracle parameter file (init.ora). Solaris will not do
async i/o correctly, and each async i/o request will take extra time. When you
do this, you may want to use multiple DB writer processes.
Finally, how are you loading the data? If you are using
SQL*Loader, it may be that the SQL is slowing you down. Direct path loading
should show a dramatic change if you are not using it already.
Just
an FYI, it is generally going to be counter productive to have table and index
data on different volumes oin the Filer. Because of the way the Filer does I/O
to the disks, it generally does not help to seperate things that way. By having
more drives in a volume, performance will improve. Although it sounds like your
problem is just getting the data out of the Oracle engine and to the Filer
(hence the direct path loading should help that).
-tom