Is there a direct correlation between data model size and report/visualization speed in the service?

[Anonymous]

Data Model & Report A: 5 tables inside of the data model which are all being used across the report on miscellaneous visuals and calculateions

Data Model & Report B: 10 tables inside of the data model. But the report itself is the same exact report  as Data Model & Report A, except there just happen to be some extra unused tables in the data model which inflates the overall size of the data model.

Will Data Model & Report A & B perform the exact same in the PBI service, or will B be slower?

I know all over the place, if people claim their report is running slow one of the most common recommendation is to clean up the data model by reducing the number of unused tables & columns, but does this actually increase performance in the service?

[jdbuchanan71]

Hello @Anonymous 

Table expansion is driven by all the relationships in your model so even if they are not used in the report, the extra tables could impact the performance.  If you apply something like CALCULATE ( [measure], FILTER ( 'Sales', filter criteria ) ) you are filtering the expanded sales table which will pick up the related but un-used tables.  This is why it is best  to avoid putting an entire table into a filter expression and instead filter on specific columns.   Here is an article talking about table expansion.

https://www.sqlbi.com/articles/expanded-tables-in-dax/