1.

Detects failures  

 2.

Restarts applications on existing or alternate system(s)

 3.

Transfers network / IP addresses transparently

 4.

Returns recovered system back into the cluster