4 Replies Latest reply on Dec 3, 2019 3:27 PM by Patrick Mcmahon

    Help understanding the workflow for created a data extract and publishing back to server

    Patrick Mcmahon

      Hi all!

       

      I have a data connection published on Tableau Server that consists of joins between DB2 and two separate Excel spreadsheets.  I've been able to publish that connection correctly and use it directly from the server.  Since the db2 table contains tens of millions of rows, it is slow as molasses therefore I want to create an extract of the data connection to use.

       

      I "think" I did that successfully yesterday (took several hours to create) and now I have a file xxxxx.hyper in my Datasources file on my local machine.

       

      I can connect to it via a new workbook and create views, etc. and it works great.

       

      However, when I go to publish the extract, from a view, using the Publish to Server dialogue on the Datasource, it does not give me the option to schedule the refresh for this extract...

       

      which I would have expected it to do, since the original data connection is a live DB2 databased + the join to the Excel spreadsheets.  I'm not expecting the spreadsheets to refresh, but I would expect that the DB2 data should be refreshed...

       

      Here is the Publish image:

       

      DataConnection1.png

       

      I "feel" like I may have done something incorrect in building the original Extract, that Tableau no longer thinks that the Extract has a relationship to the live data connection ... but that is purely speculation...

       

      Here is the Data Connection:

       

      DataConnection.png

       

      Is there some challenge / issue / consideration that I need to deal with given the nature of the data connection (a DB2 DB and two Excel spreadsheets) ???

       

      My end goal for this was simple:

       

      1. Build a data connection on the server and publish (done!)
      2. Build an extract for that data connection (maybe did incorrectly ??)
      3. Publish the extract for that data connection back to the server ( TBD )
      4. Schedule the extract for a daily refresh so that the data is always up to date (TBD )
      5. Build views (myself and others) from the server data extract..

       

      Thoughts or pointers are most appreciated!!! Thanks!!!

       

      Patrick

        • 1. Re: Help understanding the workflow for created a data extract and publishing back to server
          Phoebe Chow

          Hi Patrick!

           

          What version of Tableau Server are you using? This sounds like it could possibly be a limitation of the server role you are in. Depending on the version you're using the role names might be called something a bit different but this is a good place to start. Confirm that your assigned site role allows you to publish extracts:

           

          Set Users’ Site Roles - Tableau 

           

          Cheers!

          • 2. Re: Help understanding the workflow for created a data extract and publishing back to server
            Patrick Mcmahon

            Hi Phoebe!

             

            We are using Server version: 2019.3.0 (20193.19.0913.2225) 64-bit Windows

             

            As far as my role, it is set as "Server Administrator" with a Tableau Creator License

             

            I have tried this from both MacOS Mojave as well as on the actual Windows Server itself (Windows Server 2016 Standard)

             

            I tried it again today and even though I was able to create the extract, I am not able to publish it, except as part of the workbook that I used to create it ....  which does not allow others (or myself) to reuse that extract from the Server.  We end up publishing multiple extracts of the same datasource, one for each workbook...

             

            Thanks for any thoughts!

            • 3. Re: Help understanding the workflow for created a data extract and publishing back to server
              Peter Fakan

              Hi Patrick,

               

              My go to reference for this is Jonathan Drummey's blog - just google 'o extract where for art thou'. Probably add in 'Tableau' somewhere as well.

               

              At first glance, it sounds like you might be in .twbx/.twb live/extract lala land, particularly if you are using a .twbx viz that you have published previously with the live connection to the database. This can be quite hard to undo, but I'd look at trying by sideloading the data source using Data > New data source (and setting the connection settings appropriately to your requirements), and then replacing the data in your vizzes using Data > Replace Data Source.

               

              In advance, this can fry your aliases because in effect you are ripping the engine out and replacing it with another one, however once you are setup on extracts you should see a marked improvement in the performance of your vizzes.

               

              HTH

               

              Peter

              • 4. Re: Help understanding the workflow for created a data extract and publishing back to server
                Patrick Mcmahon

                ok, I will have a look.  I started with a blank workbook to create the viz.  In our tableau server environment, we have a project for all of the data connections, and create a single workbook for each connection, with nothing in them so that we can make updates to the connection and publish it as needed (add new calculated fields, remove fields, etc.).  That then gets "automagically" pushed to all the workbooks using that connection.

                 

                The challenge for this particular connection is it is tens / near hundreds of millions of rows and is too slow to refresh for end users, so I want to create an extract of it ... however, I don't want to have "many" extracts of the same data connection floating around since that will confuse people and potentially lead to inconsistencies ...  thus my desire to create a single extract that can be published and used by all! :-)

                 

                Off to google land!  Thanks for the pointers Peter....