Your desisions depend from couple of factors:
- how long time you have for rectore database
- how many boxes you have
- what type of disk equepment you have and may you
create shared disk array or not
If time for restore DB less than 1 min you haven't any desicions exept OPS (or ORC).
If time for restore DB less than 1 h then you can use standby database
But should be ready, that you may lost last transactions.
If time for restore DB more than 1 day you can use veritas volume manager, because if you lost 1 box VVM doesn't help you.
Here's my 2 cents worth: if purely from a DB failover, auto standby is good - just remember, you may not get ALL the data over (online redos), but it'll be a quick switchover and config/maintenance is much better now in 8i. Haven't used Oracle Parallel Failover. If from OS, we use Veritas Cluster Server, which is reliable. Also there is SUN Clustering on SUN servers and HP ServiceGuard for HP-UNIX servers that are excellent for OS failover.