Take a look at:

www.olapreport.com
&
www.johnkeeley.com

Maryam,
it's like saying 'what is the best car to drive'....it depends on what
you want to achieve.....I am a believer that the archive and underlying
dimensional models live in some form of relational
database.....especially if you have any reasonable amount of
volumes....then, if you need the features that a cube product like
analysis services brings you, by all means copy a subset of data into
the cubes and use the features you need from the cube product...

Some people try to build their BI Infrastructure entirely in cube
products....this is rarely a workable solution.

Best Regards

Peter