Star schema, when to create a dimension table?

[Arie]

Hi, 

When I have one big Excel table as a datasource, is it necessary to create a dimension table for each slicer I want to use in the report? Or can I just simply take that column from the fact table and put that into a slicer? Does this make any difference regarding the performance of the reports, when the amount of rows in the datamodel is growing?

Kind regards,

Arie

[v-jiascu-msft]

Hi @Arie,

Could you please mark the proper answer as solution?

Best Regards!
Dale

[Greg_Deckler · Accepted solution]

It is not necessary to create individual dimension tables, you can just use the columns. A lot of the old rules with MDX do not apply to Tabular. That being said, you performance question is an interesting one. Here are some links that may help:

https://social.msdn.microsoft.com/Forums/en-US/39d1ea4c-4a54-4b2b-9135-e0fa09150cc2/ssas-tabular-per...

https://social.technet.microsoft.com/wiki/contents/articles/18569.ssas-taking-the-tabular-journey.as...

[MFelix]

Hi @Arie,

According to my experience, I usually create dimension tables for columns that I use in several tables, that way I can make the relationships uch easier and simplifiying the model in term of slicers.

I have made several models and dashboard and didn't notice any majour differences between using a dimension table for slicers or the actuals column in the datasource. But it also dependes on the size of your data source.

Again this is my experience with the models I have created.

Regards,

MFelix