You can create column-partitioned single-table join indexes with the following restrictions.
- The index cannot compute aggregates.
- The index cannot be row-compressed.
- The index cannot have value ordering (but they can be row-partitioned).
Column-partitioned join indexes are designed to support requests that very selectively access a variable small subset of the columns and rows, either in predicates or as column projections.
The Optimizer can also use column-partitioned join indexes to support direct access to a column-partitioned table using a RowID join.
For more information about join index usage, see CREATE JOIN INDEX Usage Notes .