8 Replies Latest reply on Nov 5, 2018 1:56 PM by Dan Cory

    Tableau Data is heavy

    jehan siti aisyah

      Hi Community,

      I have analytics with much data available.

      The problem is during open sheet by sheet, loading is too long.

      One sheet could loading until 4 minutes.

       

      Is it possible change loading become shorter?

       

      Here I attached my .twbx 10.4.

       

      Thank you

      Jehan

        • 1. Re: Tableau Data is heavy
          Sankarmagesh Rajan

          Hi Jehan

           

          Have you tried with 2018.2. New version is good for huge data. It reads faster.

           

          Thanks

          sankar

          • 2. Re: Tableau Data is heavy
            jehan siti aisyah

            Hi Rajan,

            Not yet. Can I get that link? is it need license and upgrade also?

            • 3. Re: Tableau Data is heavy
              Rob Palmer

              So, 2018.2 will help to some extent, but I've just opened it on my version and this workbook still takes a while. Instead I have a few observations that may help:

               

              • Are you using extracts? If not then do that unless you have a specific reason not to. That will improve performance. Once you are using extracts then right click the data source and make sure you optimise the extract
              • A lot of your dimensions and measures aren't used in the worksheet. If you right click in space on the dimensions and select hide unused fields then that will remove the ones not being used. Then re-generate the extracts to decrease the width of your data.
              • I can see that your underlying data is down at the millisecond level. Do you need that, or can you pre-aggregate any of this data? I would always recommend only bringing data in at the level of detail that your end users need. Otherwise your adding more rows than necessary
              • I can see that you only have two data sources in the workbook. Has this been done for a reason? You might find it runs better having multiple data sources designed to answer specific questions. From painful experience the one data set to rule them all, ends up with workbooks that don't perform as well as they could
              • Have you done any analysis on the workbook using the performance recording feature?

               

              Hopefully that helps with somethings to start looking at.

               

              Cheers

              Rob

              • 4. Re: Tableau Data is heavy
                jehan siti aisyah

                hi Rob,

                • I'm not using extracts. Please give me your instruction
                • Understood. I will remove some data which is not used
                • how to pre-aggregate?
                • Please give me how to do
                • Not yet...

                 

                Thank you

                Jehan

                • 5. Re: Tableau Data is heavy
                  jehan siti aisyah

                  hi Rob,

                  • I'm not using extracts. Please give me your instruction
                  • Understood. I will remove some data which is not used
                  • how to pre-aggregate?
                  • Please give me how to do
                  • Not yet...

                   

                  Thank you

                  Jehan

                  • 6. Re: Tableau Data is heavy
                    Rob Palmer

                    Hello,

                     

                    So to answer your questions:

                     

                    Creating extracts

                    If you edit your data source then in the top right you'll see two radial buttons, switching this from Live to extract will generate the extracts:

                     

                     

                    Pre-aggregating data

                    You've a couple of options on this side, you can either do it in your data source so that Tableau never sees the base rows (this is what I would do), or you can do it at extract. To do it at extract you can use the aggregation option by clicking edit in the screenshot above. This presents you with the following options:

                     

                     

                    In your data the roll up dates to option will not be grayed out, and that will do the aggregation on extract for you.

                     

                    Splitting out into multiple data sets

                     

                    Unfortunately this isn't a quick win and not something that you can be shown how to do without having a consultant come in and work with you. It basically means starting again I'm afraid. The best way of doing this is to write down the questions that you are trying to answer, then you design a few data sets to answer those questions and build the graphs off of that. The idea is that each data set should contain just enough detail to answer the question it is designed for. So for example, do you need to the SPN value at every single second in the total fuel used per engine graph? I suspect not, but it's not something I can answer unfortunately.

                     

                    I would suggest that ultimately this is going to give you the biggest performance gains, but it does have the greatest cost.

                     

                    Cheers

                    Rob

                    1 of 1 people found this helpful
                    • 7. Re: Tableau Data is heavy
                      James Frick

                      I read and appreciate Robert Palmer's reply about splitting up the data sets.  I am a novice with Tableau, so his answer leads me to two other questions.

                       

                      1.  How do you handle data set splitting when there are multiple levels of data?  Couldn't end up with many data sets?  For example, we have the following hierarchy of info with record counts:

                       

                         Region - 10

                         Market - 55

                         Company - 6,000

                         Contacts - 150,000

                       

                      Our users typically analyze each month counts, by region, market and company, looking for the top companies in a specific market to focus on and then look at the contacts for those companies.  The facts number in the 60 million range for the last 3 years.  If we limit ourselves to only the last 3 months, we would have 10 million facts.  We think we need to build a summary table at the company level with summary metrics for the facts.  Our fields at the summary level are about 21 columns.

                       

                      So do we build a data set per market and company with company level summations?

                       

                      2.  Once having picked out maybe the top 5 companies for a specific region, how do we drill from a company to the individuals of that company?

                       

                      Thanks for any thoughts.

                      • 8. Re: Tableau Data is heavy
                        Dan Cory

                        Actually the problem is your data sources. You joined Sheet1 to Januari on the Unit ID column, but that is the same for every record. So Tableau is matching every record to every record and creating a huge data set.

                         

                        I think perhaps you meant to union the two tables rather than join them.

                         

                        Dan