Forum Discussion

yasbos's avatar
yasbos
Icon for Resolver II rankResolver II
3 years ago
Solved

Measures in SSAS

Hello, Team. I was told that, if my organization has SSAS server, then I should create all of the measures/calculated columns in there rather than in PBI Desktop. Is that true? I am not familiar with SSAS develpment, but I can learn it if that is true. How about the relationships model that I typically create in PBI Desktop in order for me to create all the measures where multiple tables are involved? In other words, is everything in terms of relationships, calculated columns and measures that I have been doing in PBI doable/prefreable in SSAS?

It then makes me wonder, for organizations that have SSAS, it should make no sense then to allow the development of any measures in PBI Desktop. No?

Thanks. 

  • Hi,

     

    in my opinion it depends on your data, if your volume of data is small to medium and the model isn't too complex then it would probably be better to use Power BI. If the data model is large and more complex go with SSAS.

     

    If I answered your question, please mark my post as solution, Appreciate your Kudos 👍

     

3 Replies

  • DOLEARY85's avatar
    DOLEARY85
    Icon for Resident Rockstar rankResident Rockstar

    Hi,

     

    in my opinion it depends on your data, if your volume of data is small to medium and the model isn't too complex then it would probably be better to use Power BI. If the data model is large and more complex go with SSAS.

     

    If I answered your question, please mark my post as solution, Appreciate your Kudos 👍

     

  • Thanks, DOLEARY85 . I have large and complex data and access to SSAS, so I guess we shold/have to use only SSAS.

    • DOLEARY85's avatar
      DOLEARY85
      Icon for Resident Rockstar rankResident Rockstar

      Yeah, I would use SSAS then. Power BI is basically a lightweight verion of SSAS so it has it's benefits with small/medium datasets.

       

      It's usually the cost of SSAS i've found that's a sticking point for people using it so if you already have it and a large complex dataset i would certainly opt for that for the analytic side of creating models/measures etc