I've always been forced to use raid-5 on all of our databases. For the most
part it hasn't been too bad. A couple of databases we've had to use quick
i/o to get decent performance. One server using T3s just bites. Performance
has gotten so bad they finally had to do something. The company had ordered
a new array with 6 146Gb drives that they were going to raid 5 as the
solution to the database problem. Got on a call with all the parties
involved expecting once again the database itself would be called into
question. The guy in charge of setting up the array said raid 5 is not the
way to go. He said switch the 6 146Gb disks with 12 73Gb disks and go raid
0+1 and that you have to measure by throughput not by cost/mb.
Mike
--
http://www.freelists.org/webpage/oracle-l
Received on Thu Feb 10 2005 - 13:11:45 CST
