Long running storage design

  • Hi!

    I have a cube with ~30 dimensions.  When doing storage design for

    current year partition with 75% performance gain, I face very legthy

    work.  After 10 hours I got only 7% performance gain with ~2700

    aggregations.  These aggregations were processesd in 20 minutes during

    cube processing.  Microsoft states that I can have up to 128

    dimensions per cube and up to 65535 aggregations per partition. So I

    am very far from the limits.  The fact that makes me most suprised is

    that the cube processing takes very little time, compared to storage

    design.

    What am I doing wrong?

    Thanks.

  • It sounds to me that you have some very large dimensions. Do you have any description type columns in the dimensions? It is not a good idea to hold large data columns in dimensions.

    If you do, you are better off storing them in the fact table or the dimension storage will take a long time to complete.

    It could also be that your dimensions are too large to be suitable for a star schema and you should consider a snowflake.


    ------------------------------
    The Users are always right - when I'm not wrong!

  • I have very different dimensions from two members (Account Type: Large account/Small account) to ~20000 members organized in three or even four level hierarchies (Customers, Date). My fact table has ~2700000 records and a cube built completely out of this fact table (no dimensions based on other tables) having ~15 dimensions is very problematic to design storage with more then 30% performance gain. (Gets worse if I make 30 dimensions using additional tables)

Viewing 3 posts - 1 through 2 (of 2 total)

You must be logged in to reply to this topic. Login to reply