data load implementation RRS feed

  • Question

  • Hi, We are trying rework the process on how data is stored in the Dimension and its associated Fact Tables. This is the retailer information table where the retailer places stock orders for his retail outlets across the nation and all the aggregated to the transaction day and loaded into the table. Now the requirement is to capture the order information for each individual store and we will have to change the existing data accordingly.

    For instance previously

    RetailerId TransactionDate StockUnits
    AXX768    07/14/2001        18,458 
    AXX768    01/29/2002        19,830 
    AXX768    05/13/2002        17,909 

    RetailerId StoreId TransactionDate StockUnits
    AXX768    23         07/14/2001         9,400 
    AXX768    05         07/14/2001         9,058 
    AXX768    19         01/29/2002      12,430 
    AXX768    43         01/29/2002       7,400
    AXX768    36         05/13/2002     11,805
    AXX768    29         05/13/2002       6,104

    And all the historical data has to be reloaded. What would be the efficient way with minimum change in design to redo this process.


    Monday, October 29, 2012 9:55 PM


  • I would recommend that if you do not have storage constraints, you might find value in adding the new fact table and retaining the original as well.  Otherwise, you will likely just need to drop the old table and add the new one and reload.  One other thought, if StoreID is a subset of the Retailer, you could drop the Retailer ID thus reducing your row size. 

    Regards, Steve @dataonwheels http://www.dataonwheels.com

    • Marked as answer by Maggie Luo Monday, November 5, 2012 8:40 AM
    Tuesday, October 30, 2012 6:06 PM