A big headache in SSAS, especially when dealing with a large cube, is making a change that when deployed, causes a cube to become unprocessed, forcing you to do a full process so the cube is usable. Nothing is worse than making a change without realizing it will cause the need for a full process on all the partitions of a cube that will take many hours.
For example, a process full on a dimension will cause cube/measures group/related partitions to become unavailable/unprocessed. Here are other actions that will make your cube unprocessed:
- New measure group
- New measure
- Edit measure aggregation method
- Edit dimension usage
Here are other actions that will make your dimension unprocessed (and therefore all cubes connected to the dimension become unprocessed):
- Add an attribute to a dimension
- Change the order by property of an attribute
- Edit the attribute relationship
- Add or delete dimension translation
So, make sure you understand which changes will cause an unprocessed state which makes the cube unusable so you can do it off-hours so you don’t end up getting angry phone calls that the cube is not working!