2 Replies Latest reply on May 25, 2018 7:57 AM by Isaac Kunen

    Any faster way to match data of 2 files in Tableau Prep ?

    Duy Ly

      For example, I have a list of customer number in Excel and I would like to find list of items which those customer ordered in TERADATA database. The only way I can think is that I would use INNER JOIN to match those information in Tableau Prep. However, I found that Tableau Prep runs very long time for this task. However, it would run super fast if I use customized SQL which "WHERE customer_number in ( )" but I have to copy/paste those customer number. Is there anyway that I can use something like "WHERE () IN ()" in Tableau Prep module to speed up for combining 2 files instead of using JOIN ?

        • 1. Re: Any faster way to match data of 2 files in Tableau Prep ?
          Owen Price

          I don't believe there's a way to define parameterized custom SQL using values from a secondary input at the moment. Although that said... it would potentially be quite useful.

           

          Generally speaking, I would prefer EXISTS over WHERE IN, and INNER JOIN over both of them. I don't know for sure, but I imagine that Tableau Prep has to pull a lot of data to your computer before performing the join, and that's what's taking the time, not the join itself. When you use WHERE IN, or EXISTS with explicit values, you are asking the server to do the join instead, which will be faster.

           

          Is there any way you can get the contents of your Excel file on to the server and join with it there?

          1 of 1 people found this helpful
          • 2. Re: Any faster way to match data of 2 files in Tableau Prep ?
            Isaac Kunen

            If I understand what you're looking for, it's that you want to take the (small) list of IDs from the Excel file and push them over to the Teradata so that the query can cull the large data set in situ before returning.

             

            You could hand-build a filter that we'd try to push down, but unfortunately there's not a good way to do this repeatably today. It is something we're looking at, however.

             

            Cheers,

            -Isaac

            1 of 1 people found this helpful