Summary
In this chapter we've learned a bit about the theory of data warehouse and data mart design and how it should be applied when we're using Analysis Services. We've found out that we definitely do need to have a data mart designed according to the principles of dimensional modeling, and that a star schema is preferable to a snowflake schema; we've also seen how certain common design problems such as Slowly Changing Dimensions, Junk Dimensions and Degenerate Dimensions can be solved in a way that is appropriate for Analysis Services. Last of all, we've recommended the use of a layer of simple views between the tables in the data mart and Analysis Services to allow us to perform calculations, change column names and join tables, and we've found out why it's better to do this than do the same thing in the Data Source View.
