Indexing over Multi-Valued Columns
Indexing over multi-valued columns is only performed over the first multi-valued column in the index. The first multi-valued column is expanded so that every value in the column is indexed. All other multi-valued columns are indexed only over the first value in the column. For example, an index is defined over Column A, and Column B, where both of these columns are multi-valued. In a given record, Column A has the values red, and blue, and Column B has the values 1, 2, and 3. The resulting index would give entries for red-1 and blue-1. The values in the second column, Column B, are not expanded. Note that even though every tagged column may be multi-valued only tagged columns that are explicitly flagged as multi-valued via JET_bitColumnMultiValued are expanded in this way. Also, due to the fact that primary index entries contain records, it is not permitted to have a primary index over a multi-valued column because that would result in multiple locations in the index in which the record would have to reside. For more information, see the Tagged, Fixed and Variable Columns topic.
Starting in Windows Vista, users have the option to set the JET_bitIndexCrossProduct option when the index is created with JetCreateIndex or JetCreateIndex2 using the JET_INDEXCREATE structure. When this option is set on an index, all multi-valued key columns in the index are expanded, and a cross product is created in the index for every value in those columns. The example above now yields entries for red-1, red-2, red-3, blue-1, blue-2, blue-3. Again, each key column must be explicitly declared to be multi-valued via JET_bitColumnMultiValued to be expanded in the index.