Each Analysis Services project could use multiple data sources. The traditional approach is to first build a staging relational database where you import data from various data repositories within your enterprise. Subsequently, you would build a dimensional model using a Star or Snowflake schema, as opposed to a normalized model you would typically use for a transactional database, for your data warehouse that has fact and dimension tables. Lastly, you build the Analysis Services solution using the dimensional model within the relational data source. This approach is still recommended, because it allows you to have more control over your data cleansing routines prior to building Analysis Services objects.
On the other hand, SSDT does give you the flexibility to connect to various relational databases and define necessary data structures within data source views, if you don't have the luxury of building the staging area or the star schema database. However, this flexibility...