MSAS - Processing Dimensions and Cubes

As stated earlier, maps are created when a dimension is processed. However, existing maps of a dimension are destroyed when it is processed and new maps are created. Consequently all cubes accessing the dimension will find it inaccessible and the cube will be invalid. The dimension will become accessible only when the cube is reprocessed.

When cubes contain a large number of dimensions and some of the dimensions undergo a change, reprocessing a cube can become a daunting task. However, Analysis server has a solution to this problem. It is possible to Incremental update a dimension. This process does not destroy existing maps and therefore does not invalidate cubes using the dimension.

Dimension Processing

As stated earlier, maps are created when a dimension is processed. However, existing maps of a dimension are destroyed when it is processed and new maps are created. Consequently all cubes accessing the dimension will find it inaccessible and the cube will be invalid. The dimension will become accessible only when the cube is reprocessed. When cubes contain a large number of dimensions and some of the dimensions undergo a change, reprocessing a cube can become a daunting task. However, Analysis server has a solution to this problem. It is possible to Incremental update a dimension. This process does not destroy existing maps and therefore does not invalidate cubes using the dimension.


Some of the changes that are made to a dimension during incremental update are the addition of new members, changing the property of existing members etc. Any other changes to a standard dimension will destroy the map and invalidate the cube. For example, you add new customer records to the customer dimension table. A cube that includes a shared dimension remains available to users while the dimension is incrementally updated, and the added dimension members are available in the cube after the update is complete. Because almost any property of a member can be changed by changing the appropriate field in the underlying dimension table, it is possible to have changed member names after the dimension is incrementally updated. The incremental update is also available only if a single shared dimension is updated. In other words the Incremental update processing option updates a dimension when changes have been made to the underlying tables of a dimension, but no structural changes have been made to the dimension itself.

Let us explore the screens that help the user incremental update a dimension. Let us say that we have added new sales_regions to the table and need to update the dimension to incorporate the incremented table.

Incrementally Updating a Dimension

1. In the console tree, Open Shared Dimensions folder and right click Department dimension and click Process

2. Select Incremental update option and click OK. Close the Process Log window.

3. When performing incremental update on a dimension, the Analysis server uses an SQL statement to extract information from the dimension. It is a repeat of the process followed when rebuilding the dimensions structure. However, new paths are created only for the new members. The new member is assigned the first unused number for the children under the region. For instance if Vera Cruz from Mexico is added to the dimension table, Vera Cruz will be assigned which is the first available number. However, when the dimension is rebuilt the member Vera Cruz will become the first state in the region and will be given the path.

4. Since the hierarchical view of data is important in Dimension processing, Analysis server will regenerate the Member Id for the entire dimension while incrementally updating the dimension. Changing the member Id does not invalidate the cube because the cube only looks at the path.