7 Replies Latest reply on Nov 9, 2016 6:24 AM by Chris McClellan

    Optimal work with extracts

    Irina Zholudeva

      Hello, everybody!

      I need a recommendation how to organize the work with tableau for me as analyst and for the users of my reports.

      As a source of data I use a database Microsoft SQL Server. It contains about 40 million rows now. New rows are added every night. I made 5 Tableau workbooks, which contain worksheets, dashboards and finally a story, which I present every week. Although all these workbooks use the same data source, every workbook is devoted to different topic and it would be very inconvenient for me to unite them in one workbook. Only I have Tableau license. Other users have Tableau Reader installed.

      In order to show my presentations to my users every week I open my workbooks, change the period of analysis, sometimes add some new slides, then I save extract of the data source in a folder where everybody can reach it, and then I save my workbook as a packaged workbook in the same folder. This process takes at least one hour and my computer is extremely slow during this period. And I have to repeat this operation 5 times.

      This is the typical view of my folder:

      Для вопроса по табло.JPG

      Therefore, my questions are:

      1. Is there any more elegant and simple way to work without spending more money?

      2. May be I can use one extract for 5 different packaged workbooks? If yes, how?

      3. I would be happy to get some other recomendations about optimisation of work.


      Thanks for the help in advance.

        • 1. Re: Optimal work with extracts
          Chris McClellan

          I'd be interested to see how many fields are duplicated across your 5 workbooks.  I think you could make a lot of gains by creating only 1 extract and then using that in each workbook.  So, my answers would be:


          1. Yes, 1 extract with 5 workbooks referencing the same extract. 

          2. Simply create the extract and in each workbook reference the same extract.  You can't use "hide unused fields" when you're creating the extract, but you should do that manually to figure out what fields you can remove.

          3. Can you aggregate the data at all ? ie you probably don't need 40 million rows, so aggregate the data so that the worksheets and dashboards show the same data, yet the record count is smaller and the extract will be smaller.


          Let me know if you need more help, this sounds like an easy thing to improve and get it working better for you

          • 2. Re: Optimal work with extracts
            Irina Zholudeva

            Chris, thank you very much for your answer.

            So I should creat an extract from one of my workbooks, then in other 4 workbooks I choose this extract as a data source, then I should refresh only this one extract once a week. Right? Can I make somehow the automatic refreshment of the extract once a week?

            About aggregation. My row in data base is one position from every invoice. Normally while analysis I compare week by week sales. But in some of the reports I may show the information from the exact invoice. How should I make aggregation in this case? Or may be it does not influence at all?  

            • 3. Re: Optimal work with extracts
              Chris McClellan

              No need to aggregate if you need down to the invoice amount


              You can automate the refresh a few ways, but publishing to Tableau Server and setting up there is the easiest.

              • 4. Re: Optimal work with extracts
                Irina Zholudeva

                And what are the other ways to automate the refresh?

                • 5. Re: Optimal work with extracts
                  Irina Zholudeva

                  Returning to my first question. I created extract from one of my packaged work books. I opened another packaged work book and changed the data source to that newly created extract. I clodes the former data source. The packaged work book has Live connection to it, and the icon of a data source is a single keg, not a double (which is a sign of an extract). So my last packaged work book understands this extract as an ordinary data base, not an extract and what should I do to refresh it as the button Refresh extract is not active? Should I open the first packaged work book from which I created it and do it there?

                  I am sorry for being annoying.  

                  • 6. Re: Optimal work with extracts
                    Chris McClellan

                    You could script it a few different ways on your machine, do you have Tableau Server because that's an easier "no code" solution

                    • 7. Re: Optimal work with extracts
                      Chris McClellan

                      Hmmm you're right.  I hadn't noticed that it's now a single keg :O  .... don't worry, it's still a TDE.


                      Yes, I'd use the first workbook to do it seeing that understands that it created the extract.