I need to create a single cube from about 50 separate, but identical relational databases, each representing a different client, each with a different value of ClientId. The client databases have dim tables like customer but CustomerId = 1 may exist in every client database representing a different customer for each client.
The fact tables in the client databases have up to 20 million rows.
How do I create a single cube from these? I've tried creating multiple data source views but do not know how to get more than one represented by the cube. Can I somehow use partitions to add or remove a client? Data is typically added at different times for different clients, so it would be preferable to process on a client basis.