2 Replies Latest reply on Oct 4, 2018 1:43 PM by Umut İşcan

    Join vs Blend question

    Ben Perlman



      (Working off extracts), Am I right to assume that if I am able to join the tables I need, get them to  the same date granularity, and pick what fields I use, set to extract, that it will maximize performance?  There are calcs that I need to do across the tables, so it is either through a blend (probably least efficient?) or with joins.


      Before I rebuild my workbooks via a join I'd like to confirm so I don't waste time only to not make it any more efficient.  I've read all the blogs I can find on this, but everything seems to point to "it depends".  So want to check this before I try.  Thanks!

        • 1. Re: Join vs Blend question
          Ritesh Bisht

          Generally we should avoid data blending when we can combine the two data sources outside of Tableau.

          If not possible, then we must identify at least one common variable shared by the two data sources you want to blend together.

          When possible, go for a join rather than a blend.

          If you need to combine two data sources and for whatever reason cannot manage to join the data outside of Tableau, your only option is a data blend.




          1 of 1 people found this helpful
          • 2. Re: Join vs Blend question
            Umut İşcan

            Tableau is a self service visual analytics product and users need to blend different data sources to find out answers. They must be empowered to blend different datasources over conformed dimensions. This is also the core principle of dimensional modeling : You join one fact table with dimensions to create one star schema. Star schemas are blended over conformed dimensions, they can not be directly joined because their grains are mostly different. This is the standard approach for performance and governance. But data blending never performs well, if conformed dimensions have high cardinality.

            1 of 1 people found this helpful