Real best ASM practices
The reasoning behind this comment is that all the archive/redo/controlfiles/backups will end up by default in ASM. This is great until something corrupts ASM metadata and one loses one or more disks. Presto, no backups, no recovery. Redundancy is external due to financial reasons.
It would appear that to really make ASM both work and be recoverable one should
-- allocate a second archivelog destination external to ASM (= filesystem)
-- put at least one member of each redo log group external to ASM