Sign in
 
Home20122008 R2200820052000LibraryForums
SQL Server >
> 
> 
Best practices for SQL Server Analysis Services Projects with Multiple Developers

Answered Best practices for SQL Server Analysis Services Projects with Multiple Developers

  • Saturday, January 26, 2013 5:06 AM
     
     
    Sign In to Vote
    0

    We have a scenario where multiple developers work simultaneously on the same cube, using same or different measures and dimensions. We often encounter issues when one developer overwrites other developers’ changes or when a change depending on another change is not ready for production. Both TFS and VSS have been helpful or we lack right strategy to make them work. Please any idea of best practices, strategies or tools for effective concurrent SSAS development? Thanks

    BI Developer

     

All Replies

  • Sunday, January 27, 2013 12:24 PM
     
     
    Sign In to Vote
    0

    This is  my methodology,  may help you

    • First get the DSV (data source view) ready by  assigning chunk to individuals taking and build the final DSV
    • Run the cube wizard ,identify all dimension and measure groups.
    • Your basic cube is ready now put this cube solution into source control
    • Assign dimension design to individuals.
    • Once you are done with the design take the xmla script (alter) of dimension and put it in source control (individuals)
    • Team lead would take all this dimension xmla alter scripts and executes on the ssas database
    • Update source control the latest cube solution
    • Assign measure groups to individuals
    • Do the same for measure groups
    • Now your final cube sturcture  is ready, what is pending is calculation, partitions, aggregation etc are important that key person in team can handle
     
  • Sunday, January 27, 2013 4:32 PM
    Moderator
     
     
    Sign In to Vote
    0

    Hi,

    The best way is to have very clear definitions and named standards for all cube objects and do a review of the development work before it is being published in production. You will probably have to add responsibility for each cube object like developer x owns the product dimension and developer y the time dimension.

    Technically you will have to use reviewed xmla scripts for each cube object when you build a new version of the cube-

    I would not allow concurrent work on the same cube but have a dedicated developer for each cube.

    BR

    Thomas Ivarsson

     
  • Monday, January 28, 2013 3:33 PM
     
     
    Sign In to Vote
    0

    Thanks, we are actually foloowing the steps stated here. The issue we are having is merging SSAS files from different developers against a single cube. For example , developer A checks out “Sales.cube” , add two additional attributes and an hierarchy to a dimension, then deploys the cube to test and checked the file in. Developers B checked out the cube and make changes to the measures and partitions and check in again. How can we merge the changes between the two checks in files without metadata conflicts? Our experience is that we are unable to merge these files I read that this is because,  the large sections of the file  check out by develop A will be different from copies checked out to developer B due to metadata capturing the state of Visual Studio and the local AS server. Is there a way round this or other alternative solutions?

    BI Developer

     
  • Monday, January 28, 2013 11:54 PM
     
     Answered
    Sign In to Vote
    0
    This link provide information and code to fix the issue

    BI Developer

    • Marked As Answer by Abioye Monday, January 28, 2013 11:54 PM
    •