Analysis services hang while processing the cube

[bhags]

When trying to process a cube analysis service hangs. Is this due to any specific reason

 

[MDXer]

How many aggregations are on the cube? Sometimes when you have a large number of low level aggregations theycan take a long time to processes. In my experience I have had 3 low level aggregations take close to 2 hours to complete. Is there any particualr reason that you think processing has halted?

Shoot Me! Shoot Me NOW!!!
- Daffy Duck

 

[bhags]

We have atleast 5 aggregations. This could be one of the reason. But one interesting thing we are noticing, when we are processing the cube the temp db hogs all the 25GB assigned, why is this?

 

[MDXer]

what type of measures are contained within the cube?? Sum Count or Distinct Count?

Shoot Me! Shoot Me NOW!!!
- Daffy Duck

 

[bhags]

one of them is distinct count

 

[MDXer]

THe distinct count would be the problem. Placing a distinct count in a cube with other types of measures totally messes up how the cube works for a backend standpoint.

THe best and most reccommended way to include distinct count measures in a cube is to place them into a dimensionally identical cube to the cube that has the other measures and then delete all but the distinct count measure, and delete the distinct count from the initial cube. You can then compine these bin a virtual cube. Performance will increase dramatically and the design is transparent to the user.

Shoot Me! Shoot Me NOW!!!
- Daffy Duck