According to SQL Server Books Online: Partitions can be merged only if you meet all of the following criteria: http://technet.microsoft.com/en-us/library/ms175318.aspx. They are in the same cube.
· They have the same structure.
· They have the same storage modes.
· They contain identical aggregation designs.
· They share the same string store compatibility level (applies to partitioned distinct count measure groups only).
The BOL also talks about a known behavior with “Merging Partitions That Have Different Data Slices” http://technet.microsoft.com/en-us/library/ms175410.aspx
“When you merge partitions that have data slices specified in the Partition Wizard the resulting partition can contain unexpected, incorrect data after it is processed. To prevent this, you can create a filter that specifies the data in the resulting partition.”
However, you may find that even you meet all the requirements above and you have no plan to reprocess the data after merge. You still run into some unexpected situation.
Errors in the OLAP storage engine: The slice specified for the Calendar Year attribute is incorrect.
Look like we are stuck - If we merge two existing partitions with different slices, we have a constraint issue. If we update the slices and reprocess the partitions, it defeats the purpose of “merging partitions”.
If you run into this situation, my suggestion for you is to
C S John Lam | SQL Business Intelligence | Premier Field Engineering