4 Replies Latest reply: Nov 13, 2012 9:48 AM by Messi RSS

    Make change on cube before import data - pls advice,

    Messi
      Hi all,
      I have process which export all data (all levels) to txt files, clear the data and import back

      we would like to make 2 changes on the outline once the cubes are empty
      before import data back.

      Then changes plan:
      1. Update metadata which inculding changing members on sparse dimensions and adding new dense member
      2. Changing dimension order (we have planning with essbase ) to new order.

      Will those changes might affect the import data process and create a problem?

      Thansk Guys,

      Messi
        • 1. Re: Make change on cube before import data - pls advice,
          name_and_address_withheld
          I don't think re-ordering the dimensions will make too much of a difference, do you use a load rule to load the export?

          As for "changing members on sparse dimensions" - this might be an issue depending on what you are changing. Changing a member name will cause data for the "old" member to error the load, and depending on if you use a load rule or not may be tricky to get around. If you do use a load rule then you can just map the data in the load rule from the old to the new name.

          If you are changing properties of the sparse members, that too could be an issue. For example, doing a all level export will mean that these changes may not take affect until a calc is run.

          Adding dense members will not have any affect.
          • 2. Re: Make change on cube before import data - pls advice,
            Messi
            Hey
            I'm not using rule file in this process, it's export import as is.
            I'll make the update md after importing.

            Thanks a lot for helping.
            Messi
            • 3. Re: Make change on cube before import data - pls advice,
              GlennS_3
              reordering the dimensions can affect the reload as there are possibilities to create fragmentation. For this piece of the puzzle, I suggest exporting making the changes, reimporting then export again. This will put the data file in the best order.
              The second piece of the puzzle, the renaming of members. You could do the renames by renaming the members and adding the old member names as aliases of them, then the data file can load to the aliases with no problem. An alternative would be to do the rename before exporting the data as it does not take much time to restructure that type of change, it is only an outline restructure. The one issue you can get is if you are renaming A to B and Cto D and E to A. IF you do this in one step, the outline will think you just moved A and move its data with it instead of bringing E's data. In this case you have to do a two step change doing everything except E to A. save the outline then move E to A and resave.
              • 4. Re: Make change on cube before import data - pls advice,
                Messi
                I want to work by the book ,however exporting 100GB take here 2 days, importing another day
                make this procedure again, require in total 6 days of downtime,
                maybe after the first import, I can check the fragmentation status, and then decide if it worth the downtime..
                I most worry about the data correctness, and I understand it doesn't affect
                as for sparse changes, I won't take the risk and make if after the process done.