I'm having difficulties following your logic. You're concerned about implementing HA so that if a host fails, another host can restart the workloads. However, if the ESXi box running OpenFiler fails, you lose all of your critical VMs until the Host with OpenFiler on it comes back up.
Without a shared storage solution, you might be better off implementing redundancy across your hosts on local storage. eg, put a domain controller on each host.
If you go ahead anyway, don't use RAID 5 - that's just asking for issues. Use RAID 6, especially for supporting critical VMs.
