Questions on redundancy for RAC

Question

Grid Infra version: 11.2.0.4
Platform: Oracle Linux
http://docs.oracle.com/cd/E11882_01/server.112/e18951/asmdiskgrps.htm#OSTMG13711
Question1.
We use Enterprise class storage from Hitachi Data Systems called VSP for our diskgroup LUNs. They provided frame(hardware) level RAID protection. Hitachi SAN admins confirmed that the LUNs they allocate are created our disks from different failure groups. In this case , do we need to use High or Normal redundancy ?
Question2.
We are going to build a 3 node RAC cluster for a very critical application. For the +OCR_VOTE diskgroup, we are going to use LUNs from Hitachi VSP. SAN admins are going to allocate 1GB LUNs for this. In addition to the hardware level protection provided by Hitachi , we would like to have an ASM level redundancy as well. We prefer High Redundancy for  +OCR_VOTE diskgroup. So, how many 1GB LUNs will be required for this ? How is this calculation done ? The above oracle documentation just says "three way mirroring" for High redundancy , does this mean, that,  to protect data in a 1GB LUN we need another three 1GB LUNs (total four 1GB LUNs ) for High Redundancy ?

Answer 1

Thanks. Billy.
It looks like +OCR_VOTE diskgroup is a special case in which
the number of disks for Normal redundancy is 3
the number of disks for Hig redundancy is 5
Using three 1GB LUNs , i tried to create +OCR_VOTE diskgroup with High redundancy. But, it errored out saying
INS-30510 Insufficient Number of ASM disk selected
Action:    For a disk group with redundancy level 'High', at least '5' disks are recommended.
So, I had to allocate five 1GB LUNs in total to create a High Redundancy +OCR_VOTE diskgroup.
Confirmed from ML Note ID:428681.1
For Voting disks (never use even number of voting disks):
External redundancy requires minimum of 1 voting disk (or 1 failure group)
Normal redundancy requires minimum of 3 voting disks (or 3 failure group)
High redundancy requires minimum of 5 voting disks (or 5 failure group)
Any idea why do we need 5 disks instead of 3 disks for High redundancy when it comes to OCR_VOTE diskgroup ?