The Cúram generator allows for indices other than the primary index to be created on database tables. Any number of indices can be created on each table, with the usual speed vs. database size trade-offs associated with indices.
An index for a database table is specified in the input model by adding a relationship of stereotype index between an entity class and a struct class.
The fact that a struct is being used to represent an index does not have any side-effects on the struct apart from those mentioned in the rules below, i.e. the struct can still be used as an argument to an operation. Typically the struct would be used as both a key parameter and as an index to support database accesses via this key.