Overlay datasets provide an efficient, flexible and scalable mechanism to represent the logical replication of one or more prior defined datasets. Only changes made to an entity in an overlay dataset's underlying dataset are replicated into the overlay dataset (such changes do not affect the underlying dataset). Read operations directed to the overlay dataset will find entities in the overlay dataset if they exist and in the underlying dataset(s) if no overlay-specific entity exists. Accordingly, overlay datasets provide an efficient mechanism for making changes to an existing dataset without suffering the high processing time and storage overhead associated with prior art copying and versioning techniques. Overlay datasets also provide a natural mechanism to keep two or more datasets in synchronization because changes to a base or underlying dataset's entities are “visible” in its associated overlay dataset (unless the entity has been modified in the overlay dataset).