5 Replies Latest reply: Jul 25, 2016 1:46 PM by Vinay kumar Bangari RSS

    QVD

    Deepanshu Chamoli

      Hello All

       

      It would be grateful for me if someone help me in the query of Qlik Sense

       

      I have created the multiple qvd's from different sources. When I accumulate all the qvd's in the single app; the size of the app increases to 5G.

       

      I have made certain sheets based on the qvd's.

       

      The issue I am facing is that when I click on the sheets it take lots of time to show the visualization. Its shows loading and loading. In the same manner when I used filter on it it takes the shows the bubble for 10 to 15 minutes..

       

      Thus the performance level is very very poor.

       

      Can someone please suggest how can I used my qvd in a better way so that the performance level becomes better.

       

      I need to use all those qvd in a single app... Can we save the qvd in some other extension or some thing else??

       

      Please suggest

       

      Regards

      DC

        • Re: QVD
          Vladimir Komarov

          You need to check your model.

          Sounds like you have a synthetics keys or issue with your table's joins (if you are doing it).

           

          And if everything is ok there, I would suggest to re-design your UI.

          You probably do not need to show all records at once (as I am assuming you are doing now).

          It's recommended to use some aggregation on first page of the app and drill-down (based on selections) UI on the following pages.

           

          Regards,

          Vlad

            • Re: QVD
              Deepanshu Chamoli

              Thanks Sir!!

               

              In my data model there are no synthetic keys..

               

              The qvd store more than 100 million records which might hangs all the things

               

              Just to add on this my Qlik Sense tool works on Client machine

               

              I read about the Buffer load which increases the performance level.

               

              Can we do this by Buffer load and if yes how?

               

              Please suggest

               

              Thanks!!

                • Re: QVD
                  Vladimir Komarov

                  Sounds like you need to review/update your UI/Script options...

                  Do you REALLY need 100,000,000 records to be displayed at once??

                   

                  I am assuming you can do some aggregation on the main app's page (KPI, Totals, etc) and drill down to the detail reports on the following pages (after the selection(s) has been made).

                   

                  Buffer load (I am actually not sure if it's supported in QS yet) will not help you much, IMHO.

                  It will just load a portion of your data and will require additional time/system resources to load the rest of it.

                   

                  Regards,

                  Vlad

              • Re: QVD
                Vineeth Pujari

                Appears that you have too many synthetic keys or bad joins done on date dimensions.

                • Re: QVD
                  Vinay kumar Bangari

                  if you can share us more detail about your requirement. As a user we all know we cannot analyze 100 milion records on a single chart. My suggestion for you is follow this below best practices:

                   

                  1. When extracting the data from database and creating a QVD:

                  • bring in only required columns ( for example if you have 40 columns in your table but you use only 10 feilds in front end so comment rest of the un used columns and create the QVD with required feilds ). That which said if there is chance dont use select * when pulling the data from table.
                  • bring in only required data ( for example if you have data history of 7 years but for analyzing purposes you need last 3 years of data. Then bring in only last 3 years of data to the QVD. that will reduce the size of the QVDs.
                  • Make sure that most the ETLs should be performed at QVD creation level.
                  • Check if there are any aggregations then do it qhen creating the QVDs. for example if you have data at daily transaction level or zip code level. but your front end requirement says to show totals by month or city then pre aggregate the data when creating QVD. this will reduce the QVD size.

                   

                  Data model:

                   

                  • No synthetics keys should be present. if there are any let us know we will help you to remove them
                  • No circular loops should be there.
                  • Good to have but not must: Make sure joins should be done on numbers not on text or date fields.
                  • if you need any flags create them in data model or while creating QVDs. So that way these can be used in set analysis rather than if statements.

                   

                  Front End:

                   

                  • No if statements or calculated dimensions in the front end. So calculate them in script and use them in charts.
                  • use Set analysis in expressions rather than if or class statements

                   

                   

                  Let me know if that helps you