locked
Dimension design and facts in datasourceview - loading optimization question RRS feed

  • Question

  • I always build a separate datasourceview to host my dimensions.

    When building a cube the fact tables are stored in a separate datasourceview without the dimensions referring them.

    I add the dimensions at the dimension usage tab for the appropriate cube.

    In what terms does this penaltize process performance? and  if it does penaltize..how can you host multiple cubes in datasourceviews while they refer to similar dimensions? a dimensions datasourceview can only be hosted in one file . I would intuitively put the facts per cube in separate datasourceviews

    Sunday, January 6, 2013 11:02 PM

Answers

  • I would rather suggest  to create a single DSV ,  I followed in my project by creating a single DSV by mapping all the facts and dimensions of all cubes and then i created multiple diagrams for each fact table or for each cube in DSV using Diagram Organizer section . By this there,  is no need to add dimension mapping column once again  in  Dimension Usage tab  of cube  and also it would be intuitive for the developer if we have only one DSV and have multiple diagrams for each fact table or for each cube. By this we can optimize the no of DSVs.

    ------------------------------------------------------------------------------------------

    Please mark this reply as answer if it answers  your problem.

    • Marked as answer by giovi2012 Tuesday, January 8, 2013 6:56 PM
    Monday, January 7, 2013 2:18 AM

All replies

  • I would rather suggest  to create a single DSV ,  I followed in my project by creating a single DSV by mapping all the facts and dimensions of all cubes and then i created multiple diagrams for each fact table or for each cube in DSV using Diagram Organizer section . By this there,  is no need to add dimension mapping column once again  in  Dimension Usage tab  of cube  and also it would be intuitive for the developer if we have only one DSV and have multiple diagrams for each fact table or for each cube. By this we can optimize the no of DSVs.

    ------------------------------------------------------------------------------------------

    Please mark this reply as answer if it answers  your problem.

    • Marked as answer by giovi2012 Tuesday, January 8, 2013 6:56 PM
    Monday, January 7, 2013 2:18 AM
  • Hi,

    I do not think that it has any impact on processing performance with two data source views. At the same time I do not find any great advantage to have a separated dimension dsv and  fact dsv. I Think that it is better to see the whole schema with facts and dimension tables since, like said here, you can create multiple diagrams in the dsv.

    BR

    Thomas Ivarsson

    • Proposed as answer by Elvis Long Tuesday, January 8, 2013 10:24 AM
    Monday, January 7, 2013 2:27 PM
  • hi all, I forgot the diagram organizer. You are absolutely right when using the diagram organizer you would get the best of both worlds. Thanks for the hint.
    Tuesday, January 8, 2013 6:57 PM