Managing Information Technology

104 Part I • Information Technology

company can replace the capture, transfer, and presen-
tation software modules separately when necessary.
Presentation systems are often the most volatile types of
applications, and these types of systems provide manage-
ment with business value. In addition, with modern pro-
gramming languages and system generators, business
managers can customize their own presentation and analy-
sis software to meet personal needs.

DATA SHOULD BE CAPTURED ONCE Another implication
of the separation of data from applications is that data ideally
should be captured at one source and, even when not shared
from one common database, synchronized across different
databases. It is simply too costly for an organization to cap-
ture the same data multiple times and reconcile differences
across applications. For example, not long ago, a university
discovered during a review of its application systems that 12
different systems captured a student’s home address. The
redundant data management cost was estimated at several
hundred thousand dollars per year. Thus, an IT architecture
based on application independence permits a more respon-
sive, flexible, and beneficial approach for managing the data
resource.
Figure 4.3 illustrates one way to view the data
architecture (more on the information architecture,
which includes the data architecture, in Part IV). The
data architectureof an organization should contain an
inventory of the uses of data across the business units.
The architecture should also include a plan to distribute
data to various databases to support the analysis and
presentation needs of different user groups. The same
data might be stored in multiple databases because that
is the most efficient architecture to deliver data to users.
To ensure that data are current, accurate, and synchro-
nized across the organization, however, key business
data should be captured once and transferred between
databases as needed.

THERE SHOULD BE STRICT DATA STANDARDS Because

the same and similar data are used in various application soft-

ware, data must be clearly identified and defined so that all

users know exactly what data they are manipulating. Further,

shared databases and data transfer systems require that data-

base contents be unambiguously defined and described (meta-

data). The central responsibility in managing the data resource

is to develop a clear and useful way to uniquely identify every

instance of data and to give unambiguous business meaning to

all data. For example, an organization must be able to distin-

guish data about one customer from data about another.

Furthermore, the meaning of such data as product description

and product specification must be clear and distinct.

Figure 4.5 lists the five types of data standardsthat

must be established for a business: identifier, naming,

definition, integrity rules, and usage rights. Business man-

agers, not IS managers, have the knowledge necessary to

set these standards and therefore should actively partici-

pate in the standards-setting process. Often this participa-

tion happens through the role of data steward, who is a

business manager responsible for the quality of data in a

particular subject or process area.

1.Identifier The identifier is a characteristic of a

business object or event (a data entity) that uniquely

distinguishes one instance of this entity from every

other instance. For example, an employee number is

a distinctive feature of each employee, and a unique

bill-of-lading number clearly identifies each ship-

ment. It is not uncommon to find applications in dif-

ferent units of a business using different identifiers

for the same entity. As long as there is a one-for-one

match of identifier values across the various systems,

there is not a problem, but usually there is no such

compatibility. The ideal identifier is one that is guar-

anteed to be unique and is stable for a long time. For

example, a hospital might wish to use a social secu-

rity number to identify a patient. Also, identifiers

Identifier:

Naming:

Definition:

Integrity Rule:

Usage Rights:

Unique value for each business entity

Unique name or label for each type of data

Unambiguous description for each type of data

Specification of legitimate values for a type of data

Security clearances for a type of data

FIGURE 4.5 Types of Data Standards