Whenever you change the structure of a dimension (i.e. adding a new attribute or deleting a attribute), any fact tables that are related to that dimension will drop its data. Remember that a cube aggregates the data across the dimension attributes and members
(simply speaking). So when you change something, the partitions need to be reprocessed to add the new data related to that attribute. That is why when you added the new attribute, you now need to do a FULL PROCESS on that dimension to reload the data. Whenever
you do a FULL PROCESS on a dimension, the existing fact data will be dropped and the parititions will need to be reprocessed.
Frank
