4 Replies Latest reply: Feb 16, 2013 3:34 PM by Anosh Nathaniel RSS

    Data Modelling Approach

    Shubham Kumar

      Hi, What is the Data modeling approach in two-layer architecture. Let me be specific about my job. Now I am creating a QVD Loader which will reload all the tables in one place and I will be fetching those table in QVW somewhere else. So after loading the table in QVD loader qvw itself I am getting so many synthetic keys. and the same table when in am loading in qvw application am renaming all those field as per the convenient name purpose so that syn key is not there. My question is how to approach for this, should we do it on the QVD creation time or we shouldn't bother and continue as I am doing. Need guidance for a proper frame work. Best Regards, SKumar

        • Re: Data Modelling Approach
          Johannes Sunden

          Some people like the QVD's to be exact representations of their source tables, others prefer to massage the tables as much as possible before committing them to QVD's. With a larger number of app developers and mixed requirements to be fulfilled from the source data set I guess it makes more sense to apply fewer changes in the QVD's. If certain business rules are true throughout the organization which they often are, then I see no reason not to add these into the QVD's already to simplify the process of bringing this logic into the presentation app layer.

          • Re: Data Modelling Approach

            I would extend Johannes answer by suggesting you should go with the structure that maximises code re-use and simplicity.

            If you find that you are renaming the same field in multiple seperate qvw apps (or applying business rules and calculcations in multiple locations) then you should push that into your QVD loader, or a dedicated middle layer.  Simply put, you should only rename fields once and in one location.  Otherwise you will end up with the same data having different labels in different applications, or worse, different fields with the same label in different applications.

            • Re: Data Modelling Approach

              When you are generating the QVD you can drop it immediately so that you can avoid synthetic keys!

               

              Something like below:

               

              store mytable into xyz.qvd;// Storing the data into QVD

              drop table mytable; //drop the table once the qvd is generated as it is no more required in the document