Virtaul Cube in MSAS 2005

Virtaul Cube in MSAS 2005

• Chris_tr

  Hi Vania,

  I can't comment on the relative performance, since I haven't worked with large linked objects. However, from a "best practices" standpoint, Microsoft recommends a single cube with multiple measure groups over multiple cubes with linked measure groups - you can refer to this Dec. Webcast for discusion of recommended practices:

  http://www.microsoft.com/events/EventDetails.aspx CMTYSvcSource=MSCOMMedia&Params=%7eCMTYDataSvcParams%5e%7earg+Name%3d%22ID%22+Value%3d%221032285766%22%2f%5e%7earg+Name%3d%22ProviderID%22+Value%3d%22A6B43178-497C-4225-BA42-DF595171F04C%22%2f%5e%7earg+Name%3d%22lang%22+Value%3d%22en%22%2f%5e%7earg+Name%3d%22cr%22+Value%3d%22US%22%2f%5e%7esParams%5e%7e%2fsParams%5e%7e%2fCMTYDataSvcParams%5e

  >>

  TechNet Webcast: SQL Server 2005 Analysis Services: Preparing to Migrate

  >>

   

  Another thing to keep in mind is the increased complexity of maintaining  cube dimension usage with linked measure groups - all changes must be made in the cube which is the source of the measure group, as discussed in the OLAP newsgroup.

  
  
  

• FranciscoDaniel

  The AS 2005 cube, along with the cube perspectives feature, addresses some of the same issues as virtual cubes did in AS 2000:

  http://www.microsoft.com/technet/prodtechnol/sql/2005/asmigrtn.mspx

  >>

  Analysis Services 2005 Migration

  ...

  In Analysis Services 2000, you would create two cubes and then combine them in a virtual cube. In Analysis Services 2005, you have a single cube with multiple measure groups (one for each fact table).

  ...

  To add measures of multiple granularity or to view many dimensions from different cubes, you could create a virtual cube to combine multiple base cubes to give the appearance of something much larger. This allowed you to build up to the final result. This not only allowed us to deal with granularity issues, but it also helped with the memory consumption in Analysis Services 2000.

  This has changed with the UDM. Now the cube can have hundreds of attribute hierarchies, user-defined hierarchies, and multiple measure groups (which come from different fact tables). Because of this, cube design is very flexible. The problem that remains is the user context. At some point you need to narrow the view of the data so that the end user can make sense out of it. The view of the data should easily allow the user to meet their project requirements.

  Perspectives provide context for the end user. A perspective is a set of attributes, user-defined hierarchies, actions, and measure groups that you want to group as a logical collection. The perspective provides the basis for the analytics and provides context for the user. It will be very common to have a large cube with hundreds of attributes and many measures. Then many perspectives will be created for the users to interact with the collection of data that makes sense for their task.

  >>

  
  
  

• nadertawil

  Dear Deepak,

  for my use it sometimes seems more suitable the use of Linked Objects, they are very similar to the AS 2000 virtual cubes, aren't they

  What can you say about them Are their performance worse than the one that can be achieved with the multiple measure group

  Thank you very much.

  Vania

  
  
  
• Virtaul Cube in MSAS 2005