I'm not sure if I understand your question...
A view is based on tables, if the tables change, then the view content
change too !!!

Do you talk about SQL based views (views created in the database) or DSV
named queries?
In both cases the result is near the same, but using views coming from the
database allow you to change your business logic without impacting the cubes
(for example, you decide to split or merge tables, your change the view used
by the cube and the cube continue to works correctly)

Also views don't contain indexes, but the tables behind the view generally
use indexes.
In SQL Server 2005 Enterprise you can create indexed views, but if its for
load a cube, indexed views will not really helps you (the time taken to
create the index in this view is an overhead, specially if the view is
designed to fill a cube)
Indexed views are great for grouping queries like "select col1, col2,
count(*) from tables... group by col1, col2"

Remember that a MOLAP cube (default type of cube) will read everything from
the database and store it locally on the SSAS server. so the source database
is not read anymore after you have done your process. So optimizing the
database don'T help the user access but can help the processing time.

"MIB" <MIB (AT) discussions (DOT) microsoft.com> wrote