If I am not wrong, failover clustering assumes the backup
database to be a replica of the original database. Does
failover clustering itself perform the data

You are kinda wrong (sorry). During a failover, the SQL
Server instance goes through a stop on one node, and a
start on the other. In that process, the cluster
resources associated with SQL Server come online. So you
are working with the DB itself, not a copy. You are
always transactionally consistent and current to the point
of failure. Because it is a stop/start, it goes through
the transaction log for each DB and rolls back any
incomplete transactions.

Hope that helps.
[quoted text, click to view]