Single generic or multiple specific columns for measures?

[fabiuzz]

Hi guys,
I need your advice, please.
My doubt is: in terms of performances and responsiveness of PBI reports, is it better to opt for a data model design with just one (or a few) generic column(s), containing mixed but 'dense' numeric data (e.g. budget / actual values)
and then to leverage DAX expressions, creating a specific filter context for each measure, or, instead, to opt for a data model design with multiple specific numeric columns, each containing homogeneous but 'sparse' data,
pre-calculated by DB (or M) queries?
In the latter hypothesis, are null and 0s equivalent, always in terms of the engine performances, in case of summing DAX measures?

Thanks,
Alessandro

[v-chuncz-msft · Accepted solution]

@fabiuzz ,

You may take a look at https://www.sqlbi.com/topics/optimization/.