The Politics of Entity Identifiers - High-Performance Web Databases: Design, Development, and Deployment - page 314

tion work-arounds, the performance problems and inability to perform
such operations as joins and data element concatenation may still exist.
SUMMARY
Although entity identifier problems frequently exist, steps can be taken
to ensure that these problems are eliminated whenever possible. Volatile
entity identifiers can be prevented if the organization performs a stability
analysis of the data model. It must also be determined whether any busi-
ness changes, however unlikely, may result in an unstable, non-unique, or
nun entity identifier. The organization must also realize whether any entity
identifier data elements may become non-key. If these possibilities do exist,
then an entity identifier must be chosen that can retain its entity identifier
properties over time (e.g., a surrogate key). Although conducting a stabil-
ity analysis does not guarantee that volatile entity identifiers do not occur,
it lessens their likelihood.
Denormalized and all-purpose entity identifiers often occur when the
organization does not have a thorough understanding of the business infor-
mation and cannot spot denormalized or layered data elements. Following
the steps to prevent denormalized entity identifiers that were mentioned
earlier should prevent this. Denormalized, all-purpose, and parochial
entity identifiers may occur when the administrator focuses solely at an
application level during data model development. It should always be
assumed that an application-specific data model is eventually consolidated
into the enterprise data model. AU data elements should be defined at the
atomic level and all principal entities should be included. The data admin-
istrator should review the application data model and, specifically, the
entity identifiers to uncover any application-specific bias that may mani-
fest itself as denormalized or all-purpose entity identifiers.
The best way to avoid entity identifier problems and to ensure that entity
identifiers retain the properties necessary to implement correct, consis-
tent, flexible, and shared databases is to thoroughly understand the organi-
zation's business information requirements and to model them correctly.
The data base designer is then in a better position to make intelligent deci-
sions about which solution to choose if an entity identifier problem exists.