you should get used to the documentation and possibly redefine your requirements. Or at least make them explicit.
There are obviously several possibilities to get a failover scenario. Are you thinkign about the full domain or some services or some servers?
However, get used to the UCS architecture it might help you in understanding.
UCS is a “single-master mutli-dc” system. This means the master is the one responsible for any writing actions (ie password changes) while the other servers have only read-only copies.
This means if the master fails most of services located on the other servers are still accessible.
If you want to have nearly 100% really high available you should consider using something like the HA feature of VMware.
However, tell us what you exaclty want and we might give you further information.