ABSTRACT
Entity identifiers have far-reaching implications for both the data model and physical data base design. A factless entity identifier is likely to be unique, explicit, definite, consistent, and stable. A volatile entity identifier is one whose value is subject to change, become null, or that comprises potentially non-key data elements. Non-key entity identifier data elements affect the identifier’s stability, consistency, and explicitness. The data administrator must update the data model and repository entries to reflect the data element’s removal from the entity identifier. Denormalized entity identifiers also make it difficult to define indexes for the physical database designer. The administrator must ensure that the physical domain of the foreign key matches that of the entity identifier to maintain referential integrity and enforce logical domains. Surrogate keys are often proposed as an alternative when problems with entity identifiers occur. An alternative to dealing with denormalized and all-purpose entity identifiers is to introduce redundant business information.
