>>>
>>>Kevin,
>>>
>>>Thanks for correcting me "a 4MB stripe width will reduce the
>>>odds there will be cross-stripe reads, but in no way
>>>eliminates it.", you're quite right.
>>>I should have made that point myself.

no worries...I'm not sure people are yet taking a
sufficiently simplistic look at SAME. It really is supposed
to be simple... threads bled off into confusion between
stripe width and depth and I sort of bailed. The
simplest rule is stripe across all disks. The only
subtlety is regarding what st widths some arrays support
and whether you have to use a software vol mgr to
cunk together LUNs from the array. Oracle can do a 1MB
IO (scattered read, direct path read/write) so make the
stripe width larger than that...and if the array wont
stripe that large (as some I work with), then go with the
largest it does support.

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