We have several SSAS cubes with shared dimensions. At the end of our ETL , we want to process the cubes.
To date, we use a single Analysis Service Processing Task (in SSIS) for this. The problem we found is that if one cube fails, they all fail.
We want to split the cubes to individual tasks. Can they be run in parallel or will we experience locks (and thus should be run in serial)?
One SSAS database with many cubes or multiple databases each with one cube? If one database, I don't think you will achieve parallel processing of multiple cubes if they are not in the same transactional processing batch (i.e. one AS processing task). You may think they run in parallel but behind the scenes I believe they will be serialized. Watch in profiler to confirm that.
Maybe you should start a different thread to get advice on resolving the errors?
