9 Replies Latest reply on Feb 11, 2019 6:44 AM by Donna Coles

    Using Published Extracts

    James Frick

      I have a dashboard with 6 embedded custom SQL queries as data sources.  It is tedious to work with and slow to refresh.

       

      So I created 6 local dashboards, each with a different customer SQL query from above.  I then published the 6 data sources as separate data sources on the server.

       

      Now I am replacing the original 6 embedded data sources with the published data sources in the dashboard.

       

      I have a question though:  when I went through the process of adding a data source and replacing the old data source, and then look at the data source tab, it shows the options "LIVE" and "EXTRACT" for each of the data sources.  What is confusing to me is that the data sources I am pointing to are Extracts already.

       

      So what do the "LIVE" and "EXTRACT" for each of the data sources on the dashboard mean?

       

      James

        • 1. Re: Using Published Extracts
          Jeff Strauss

          Hi James.  I've seen this terminology before and admittedly it does seem a little misleading, but here's my interpretation.

           

          The workbook that points at published datasources will show the connection as "LIVE", you can think of this as hitting a database which in this case happens to be a published datasource.

           

          If you click on the published datasource it should show the connection as "EXTRACT".

          • 2. Re: Using Published Extracts
            Michael Gillespie

            This is one of those "literally true but really unhelpful" things. 

             

            I always think of it as "a LIVE connection to an EXTRACT data source".  Which is true, but kind of meaningless...

            • 3. Re: Using Published Extracts
              James Frick

              Are you saying my dashboard will have a Live connection to a published data source which is an extract?

               

              Does that mean at the point I see this, I should not worry whether I pick Live or Extract?

               

              Is either option better or more advantageous than the other?

               

              Thanks for the reply.

               

              James

              • 4. Re: Using Published Extracts
                Donna Coles

                Hi James

                Here's a thread I commented on with explanations of the the live/extract label on the UI. Extracted Data Source showing up as Live connection?

                 

                I'm not entirely sure what you mean by 'picking' live or extract... I'm not clear how you get to choose...?

                 

                An extract means the data is being stored within the Tableau's in memory data engine on the server and so can be more performant than connections which are live and so require any queries to be run on the SQL Sever / against the excel spreadsheet etc etc.

                 

                Donna

                1 of 1 people found this helpful
                • 5. Re: Using Published Extracts
                  James Frick

                  I understand the difference between a Live connection and an Extract.  What I don't understand is why we see this at the top right of the screen on a list of published data sources:

                  If my published data source is already an Extract, why are we given the ability to pick either Live or Extract?

                   

                  Having read your post, let me write what I think you are saying:

                   

                  1.  I can create a data source that is an extract of the data.

                  2.  If I publish the workbook with this, the data is in the workbook and the workbook needs to be refreshed every morning.

                  3.  If I publish the data source first, it has the extract of the data (and is refreshed on schedule)

                  4.  If I use published data sources in the workbook and then publish the workbook, with the Live option checked, the workbook connects to the published data source, i.e. the workbook does not contain the data, but the published data source contains the data.

                  5.  If I publish the workbook with the Extract option chosen above, then the Workbook pulls the data from the published data source and saves it in the Workbook, which for normal dashboard options does not sound like a good idea, i.e. I now have two copies of the data, one in the published data source and one in the workbook, and I have to worry about refreshing both of them.

                   

                  Is this correct?

                   

                  Thanks for the post,

                   

                  James

                  • 6. Re: Using Published Extracts
                    Donna Coles

                    Hi James

                     

                    In Tableau Desktop, if I connect to a published data source on my Tableau Server, I don't see the Connection option you refer to above. The decision to be live or extract is part of the definition of the already published data source.

                     

                     

                    I only see the options you refer to if I create a connection to SQL Server/ Excel/ text file etc and the option is basically saying whether you want to keep that connection live to the data source or pull the data into the workbook....

                    It might be that my approach to connecting and developing is different from yours and so I'm seeing things slightly differently...

                     

                    I think your confusion (and mine) is around the ability to publish a data source to be a server data source at the same time as publishingyour workbook.  I'm struggling to understand what you mean by point 4 & 5 above, as in my mind, if you're using a published data source you don't have the live/extract radio button to select..

                     

                    I tend to develop published data sources independently of the workbook, probably because the options to do both at the same time didn't exist in the version of Desktop I started with, so its just habit.... ie

                    1. open desktop and connect to say SQL Server and build data source / add calc fields / set default format for measures etc

                    2. decide whether data source will be live or extract.  If extract, create extract (usually I do this by right clicking on data source and selecting Extract Data - you'll be promoted to save extract somewhere; you get same behaviour by selecting the extract radio button on the data source screen)

                    3. publish data source to server.  I do this by right clicking on data source and selecting 'publish to server'.  If its an extract i'm prompted to select a refresh schedule time.

                    4. data source will now be published on the Tableau Server and will be listed as live or extract depending on decision made.

                    5. now develop the workbook by opening a new workbook and connect to data source on Tableau Server as per pictures above, selecting the data source that was just published from the list presented.

                    6. depending on what's required, I may have a need to extract this data source too (eg for performance reasons), which I do by again right clicking data source -> extract data.

                    7. publish workbook to server (server menu -> publish workbook).  If the data source has been re-extracted again into the workbook, I'll need to select a refresh schedule time.  If the published data source I selected was already an extract, then I need to make sure the timing of the workbook refresh is after that of the data source refresh.  In this scenario then yes, the server contains a copy of the data and the workbook contains another copy of the data, but there are valid scenarios when this is necessary.

                     

                    Hope this helps with your understanding.

                     

                    Donna

                    1 of 1 people found this helpful
                    • 7. Re: Using Published Extracts
                      Michael Hesser

                      Hey James-- let me jump in here and add a little to what others have already said (please follow their links, as what I'm about to say may duplicate what they've already indicated).

                       

                      This is how I think of it:

                      Suppose we have a lengthy shopping list and we're asked "Would you like to use the list as-is" vs. "Would you like me to create a new list where everything is in order?"

                       

                      We now that a shopping list that is in order is more efficient, so most of the time we want to go with option #2 (even though it takes some time going through every item and confirming it is in place). But the one time we don't want to do this is when we know the list is already organized.

                       

                      This is much the same way with Tableau extracts live vs. extract: if we're already dealing with an extract (an optimized file), we'll just be creating an extract-of-an-extract (likely a copy of the same file).

                       

                      So in those cases where you're referring to an existing extract created by external Tableau process, you should refer to the LIVE version of the extract; otherwise, you run the risk creating an extract of the extract, which can quickly bloat your data sources.

                       

                      Consider we start with (1) a simple XLS file.

                      Creating a extract from it produces a HYPER extract file (2) which we can refer to in our original viz.

                      Referencing this LIVE (3) allows us to us it in other Tableau programs. Note that it does not create additional files-- it refers only to the original extract.

                      ...but creating an extract of this already optimized file creates a NEW file that is likely the same (4).

                       

                      We probably do not need both States.hyper (Extract) and States.hyper (Extract) (Extract).

                       

                      Hope this helps?

                      1 of 1 people found this helpful
                      • 8. Re: Using Published Extracts
                        James Frick

                        I think I understand now.  Thanks to everyone who responded.  I really appreciate your efforts.

                         

                        For Donna, just so she can she what I am seeing, I offer the details below.  I am using v2018.1.7.

                         

                        Here is what I see if I want to create a new workbook based on published data sources and I select Data > New Data source.  I did not make the image big enough, but under the Data Source "Call List", you would see my other published data sources.

                        (The blue rectangle is my addition to blank out the server name.)

                         

                        If I pick the published data source named "Call List", here is the next screen I see:

                        I have added the red box to show the option where I can choose Live vs Extract.  I understand now that the published data source can be a Live Connection or an extract.  (We use only extracts as we have very large data tables.)  Also I understand that the above choice in red applies to the worbook I am building.  Thus I want a Live Connection above to a published extract.

                         

                        Again, I appreciate everyone's input, and I appreciate how active and effective this forum has been in helping me understand Tableau.

                         

                        James

                        1 of 1 people found this helpful
                        • 9. Re: Using Published Extracts
                          Donna Coles

                          Thanks James!  Just tried with Desktop v2018.3 and I see what you mean now -  doh!  :-)

                           

                          Glad you're sorted now!

                          Donna