Thank you all! Your answers do help me in finding some light inside the
tunnel, and believe that most of my questions have been answered. Having kind
of incomplete knowledge on the matter (and not having english as a mother
tongue), I can say that again and again I find myself having questions in
this whole subject. Kevin's and Russ' posts were really helpful in this. So:

a. Clustering requires Windows 2003 Clustering first.
b. Servers must be identical.
c. Clustering does not guarantee a non stop work of the server (if it is
considered as a unit), but rather an immediate undertaking of new processes
that need the server at that time. If I understand well, processes from
outside the servers clustered will appear to have a slight timeout, and
immediately will be resolved afterwards by the failover server, while
processes running in the main node, will not continue, unless if there are
automated procedures that will repeat them in the failover node after a
failover.

Am I right?

Again I want to thank you for your kindness and patience with me. I do feel
well to find a place so helpful.