4 Replies Latest reply on Aug 19, 2019 4:32 PM by Vikas Panwar

    CDC on selective columns Golden Gate

    laknar

      how to do delta capture(cdc) on selective columns using OGG. We only need to trigger the changes on required columns for downstream applications(especially text attributes).

        • 1. Re: CDC on selective columns Golden Gate
          K.Gan

          Specify the columns you need in the MAP (much easier to do rather than from Capture, ie extract), so if you need col1 and col2 from tablea, just do

          MAP SCHEMA.TABLEA, TARGET SCHEMA.TABLEA, COLMAP (COL1=COL1, COL2=COL2);

          In fact the target table need not even need to be the same as source. Example if tablea has 20 cols and tableb (your target) has 2 cols with a completely different name you can do

          MAP SCHEMA.TABLEA, TARGET SCHEMA.TABLEB, COLMAP (COLA=COL1, COLB=COL2);

          The take up target columns also need not be the same size or type but you will need the OGG conversion functions. The above example assumes the same data structure.

          Cheers

          Kee Gan

          • 2. Re: CDC on selective columns Golden Gate
            laknar

            OGG is generating the target as smaller files. As and when there is a update happens in source OGG will generate the target files(file name will have time stamp appended). how can we define these target files as target table. Please advise.

            • 3. Re: CDC on selective columns Golden Gate
              K.Gan

              I don't understand this "OGG is generating the target as smaller files.". OGG does not do that unless you replicate DDLs or get extract to execute a script.

              Maybe you can show me what you actually do with your parameter and script.

              Cheers

              Kee Gan

              • 4. Re: CDC on selective columns Golden Gate
                Vikas Panwar

                Hi K.Gan,

                 

                In the approach you suggested consider below scenario

                 

                table having 5 columns

                col1 col2 col3 col4 col5

                 

                user want that col3 and col4 needs to be considered for CDC

                 

                with the approach (map columns on target side) you suggested it will bring the data for col3 col4 ALWAYS even if there is any change for that record irrespective of what column is being updated on source side.

                sometimes that is applicable if transaction volume is low.

                 

                if we specifically mention column names on extract side -- will that resolve these extra transactions to flow to replicat ?

                 

                 

                laknar  -- I am not saying that Kee suggested wrong thing but this is additional you need to think of

                off-course Kee can correct me, if my understanding is wrong.

                 

                 

                Mark this post helpful, correct or like it appropriately. This will help others

                 

                -VP