Should I use the dimension tables in the companies db or just ref/dupl fact tables

[akhaliq7]

In cases where I am bringing in dimension tables such as the customers table which may have 200,000 rows which most of them being historic data should I instead just reference/duplicate my fact table and use the customer id and customer name from my fact table instead as the fact tables have less rows once I am done with the filtering than my dimension tables. 

[akhaliq7]

Thanks that is similar to what I currently do which is to inner join the fact table to the dimension table to get a smaller dataset but wanted to post the question to get more ideas and advice. Yes the fact table has like for one report only 20,000 rows and the dimension tables because they contain historic data are like 200,000 rows long. Such as the customers table therefore it is not user friendly or good for performance to bring in a table with 200,000 rows and use it as a dim table

[ibarrau · Accepted solution]

Hi. I'm not sure I'm following the question, let's see if this helps. The engine behind Power Bi will perform better with star schema. If you are thinking about just importing fact without dimensions the performance for filtering a name of the customer directly from fact instead of dimension won't be the best.

If you are saying that the fact has less than 200,000 customers and you could just take those to became the dimension, you might be able to do it like that yes, but I would recomend bringing the customer table dimension left joining the fact just to let you filter by "customers with atleast one row on the fact". That would be a smaller customer dimension with all the columns.

I hope that helps,