Different Named Queries for Partitions
To base multiple partitions for a measure group on a single table, you can create separate named queries in a data source view that return the different rows required for each partition. You can then base a separate partition on each of the named queries.
When you use different named queries, ensure that no data is duplicated among the partitions. For example, if one named query returns transactions for 1997 only and another named query returns transactions for 1998 only, these tables contain independent data. Similarly, named queries for distinct product lines or distinct geographical areas are independent.
It is possible, but not recommended, to use different named queries that contain duplicated data. In this case, you must use filters in the partitions to ensure that data used by one partition is not used by any other partition.
When you create mutually exclusive named queries for partitions in a data source view, ensure that the combined data for the partitions includes all data from a measure group that you want to include in the cube. Make sure that you do not leave a default partition based on the entire table for the measure group, or else the query based partitions will overlap the query based on the complete table.
Specifying a named query for a partition is no different than specifying a fact table. You specify a named query on the Specify Source Information page of the Partition Wizard. On this page, next to Look in under Partition Source, specify a data source view in which to look, next click Find Tables, and finally, under Available Tables, select the named query that you want to use.
For more information about creating named queries, see Replacing a Table or a Named Query in a Data Source View.