>Can anyone please explain me main differences between relational DB and warehouse (Point me to web site)? <<
Get a copy of "Analytics and OLAP in SQL"; it is a quick overview for
the SQL programmer who has been working on OLTP applications and needs
to learn the major concepts and buzz words.
>Apart from theoretical differences I would like to know how Warehouse DB is updated? <<
In bulk, with special ETL tools if you are lucky.
>If data needs to remain unchanged then what is happening for instance if in source relational DB address for a customer will change and record gets the update. How this is loaded to DB ? <<
A warehouse holds history, not the current data. Once the data goes
into the warehouse, it stays there until it ages out, perhaps to
archives. There is also a good chance that you cannot see each
customer. They might be consolidated into higher level groups ("men,
age 50-55 years, chews tobacco" or some such thing).
>Is there any case study of SQL S2005 implementation of Warehouse, maybe tutorial that I can follow ? <<
Look for articles on the Ralph Kimball website. Frankly, SQL Server
is not the choice for a real Data Warehouse. Look at Teradata, SAND,
WX2 and other products that are built for this kind of work. Even DB2
and Oracle are also better choices. DB2, for example, has an
optimizer that detects Star schemas and generates special access
methods and joins for them.