6 Replies Latest reply on Jul 29, 2014 6:00 PM by Christian Berg

    RPD Design Questions

    Sariel-SW

      Dear All,

       

      I have made the RPD with 2 facts and 3 dimensions as

       

      Dimension 1 -->Fact 1<-- Dimension 2

      Dimension 2 -->Fact 2

       

      (The join is done on Physical Layer)

       

      When creating the analysis in Web, putting Dimension 1 and Dimension 2 together with both facts, only Fact 2 will have figures. I think the reason behind should be Dimension 2 is no linkage with Fact 1

       

      However, we would like to show the Fact 1 figure with both Dimension 1 and 2, is it possible to do so? Or I should move the join from Physcial Layer to BMM Layer?

        • 1. Re: RPD Design Questions
          Rajesh Gurram - BI

          You need to create a complex join with non conformed dimension and the fact with 1=1 and set the logical level of the measure to total for a non conformed dimension.

           

          You can refer: https://s3.amazonaws.com/rmc_docs/biforum2011/Mcquigg_Metadata.pdf

           

          Pls mark if correct/helpful.

          1 person found this helpful
          • 2. Re: RPD Design Questions
            Christian Berg

            Rajesh Gurram - BI - are you serious? 1=1 is the last option you should think about since it's the worst solution conceivable for a simple and straight-forward question like the one the OP asked.

             

            LTS content levels is the way to go for. Honestly, if you reference Jeffs presentation, then make sure that you read farther than just slide #12 which he immediately points out as the first *option* and on slide #13 explicitly puts *CAUTION*.

             

            Go one slide further to slide #14 and he presents the proper solution of LTS content levels! In all seriousness, such destructive "advice" and then begging for points gets on my nerves and is more than counterproductive since it leads the OP off on a completely wrong tangent!

             

            Sariel-SW please do the intelligent thing and model this properly in your RPD.

            • 3. Re: RPD Design Questions
              Sariel-SW

              Thanks both.

               

              Christian, you are right. I am just using the second option from Jeff's PPT (#14). I think the first one will have some unexpected results...

              • 4. Re: RPD Design Questions
                Christian Berg

                Rather "highly expected" results like forcing a cartesian product across the two facts involved. In a production system with multi-million (or -billion) fact tables that's about as bad as it gets.

                • 5. Re: RPD Design Questions
                  Rajesh Gurram - BI

                  Christian Berg,

                   

                           Thanks for correcting and pointing to the best solution.

                   

                  Your said:

                  " In all seriousness, such destructive "advice" and then begging for points gets on my nerves and is more than counterproductive since it leads the OP off on a completely wrong tangent!"

                   

                  Hope you are aware that many people who post questions in this forum don't come back and update if their problem was resolved/helpful. It's just a reminder to them to come back to the question and mark "IF" the solution provided helped them. You don't have to be so rude in responding to other's response.

                  • 6. Re: RPD Design Questions
                    Christian Berg

                    It's easy. Don't propose outrageously bad solutions like this without explanation and I won't call you out. Period.