Unless I'm missing something, FTT (Failures to tolerate) basically pretty straight forward:
FTT=1 -> one failure can be tolerated
FTT=2 -> two failures can be tolerated
Here is a easy to read table explaining FTM:
For more reading:
You need to look at it from an OBJECT or VM point of view. If you have 5 hosts and an object with RAID-5 it could be that it is a standard RAID-5 deployment, meaning 3+1 (3 data + 1 parity) and that they are spread across 4 hosts as shown in the diagram.
if the above is the case and host 5 and host 4 go down, then there's no impact to the data of this particular RAID-5 configuration. This is because host 5 is not containing any components of the RAID-5 set and as such the Object is only experiencing 1 failure, while your cluster is experiencing 2 failures.
So whenever you use 5 or 6 nodes it could indeed be that 2 hosts go down and VMs are not impacted, it could also however be that VMs are impacted. As in the above scenario if 2 hosts would go down then your FTT=1 configuration with RAID-5 is experiencing 2 failures.
Thanks so much Depping and everyone.