ASM

I am planning the migration of my data warehouse to new blade servers with 10g on Linux. I have a san with forty 72 gig drives. I am confused about how to setup the fault tolerance with the ASM. Do I setup raid at the hardware level and then use ASM or should I just give ASM all the disks and use software level raid?  

Also, when creating disk groups with ASM, should I create one group for all 100 gigs of data and another group for indexes? Do I want 1 large group or several smaller ones?  

Also, if I have 4 blade servers, is there one ASM instance on the SAN or does each blade need a local ASM instance?  

Thanks

Randy  

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