Databases Reference
In-Depth Information
Chapter 19
The Politics of Entity
Identifiers
Deborah L. Brooks
E
-
NTITY
IDENTIFIERS
HAVE
FAR
REACHING
IMPLICATIONS
FOR
BOTH
THE
.
DATA
If an organization chooses
an unsuccessful entity identifier, the results may be an inaccurate data
model, data base design and application programming complexity, poor
application performance, and the inability to establish a shared data envi-
ronment. How can one distinguish a good entity identifier from a bad one
and what should be done when one discovers an entity identifier problem?
Before these questions can be addressed, the properties that the data ele-
ment should possess or the data element sets that are being considered for
the entity identifier must be reviewed.
MODEL
AND
PHYSICAL
DATA
BASE
DESIGN
ENTITY IDENTIFIER PROPERTIES
An entity identifier should be unique to be able to distinguish one entity
occurrence from another. This is not possible if the entity does not contain
a unique value for each occurrence. An entity identifier should also be
explicit so that a value exists for every entity occurrence that is known
when it is created, definite so that its initial value is not null, consistent so
that each occurrence is associated with only one value of its identifier, and
stable so that its value never changes or becomes null as long as that
occurrence exists. An entity identifier must also be factless so that it iden-
tifies but does not describe the entity occurrence. A factless entity identi-
fier is likely to be unique, explicit, definite, consistent, and stable.
In addition to these fundamental properties of an entity identifier, end-
users should be able to access and understand the data elements and the
elements should conform to any data base management system (DBMS) or
installation-specific limitations. Whenever possible, an entity identifier
should also be able to be controlled so that an identifier with all the correct
properties can be assigned.
Search WWH ::
Custom Search