5 FAH-5 H-300
DATA ADMINISTRATION
5 FAH-5 H-310
DATA MANAGEMENT SERVICES
(CT:ITS-9; 01-30-2019)
(Office of Origin: IRM/BMP/GRP/GP)
5 FAH-5 H-311 GENERAL
(CT:ITS-9; 01-30-2019)
a. The Data Administration (DA) program is identified
by 5 FAM 600 as providing policy, program direction, and standards for
Department-wide data to be used in Information Technology (IT) development,
integration, and modification projects. This resource management function for
the Departments investment in data helps ensure compliance with industrys
best practices while maintaining an oversight role on existing systems. The
program office, IRM/OPS/SIO/EPI/DM, may be
contacted by phone at (703) 875-4400. Additional information about data
administration is also available from the OpenNet.
b. The data administration program fulfills this role
with a number of activities grouped under three primary functions: service,
standardization and supply. Data administration works hand-in-hand with
development and integration activities to provide guidance on data management
and data standards. It uses the knowledge gained in that service function to
support data standardization, based on the actual business use of data in the
Department. The program also serves as a coordinator of data sources and a
provider of authoritative data.
c. Administrative costs of the data administration
program (including maintenance of the enterprise data model, evaluation of the
conclusions of the program through the Data Administration Working Group
(DAWG), and technical support to the metadatabase integrated tool set) are
funded through the program offices. Fund citations are included in the
project-funding request.
5 FAH-5 H-312 PROJECT SERVICES
(CT:ITS-9; 01-30-2019)
The data administration program works hand-in-hand with
development and integration initiatives to provide data management expertise in
several areas. This provides immediate and continuing benefit to the Department
in accomplishing specific goals. It also provides credibility to the programs
standards and policies, as they all emerge directly from actual data use in the
Department.
5 FAH-5 Table H-312(1) Process Modeling
|
ACTIVITY
|
PURPOSE
|
|
Process Modeling
|
a. Process modeling is an analysis tool supporting
requirement identification. Facilitated conversations with employees provide
answers to questions critical to the understanding of the environment and the
purpose for the projected system. The product is the process modela
graphical image of the business processa diagram indicating the start, the
steps and the completion of the activity. This diagram serves as a focus for
discussion as the process is validated. The diagram can also identify
bottlenecks in the process, repeated steps and other inefficiencies.
b. This analysis leads to a set of statements that
articulate a desired future statethings that need to change to improve the
business process of the office. The answers to these questions point the way
to the process, data and other requirements for the new system.
c. One of the ways projects go wrong is for the
answer to be provided before the question has been asked. In 5 FAM 600, it
states that requirements are clearly, unambiguously identified before
acquisition and/or development begins. Such effort ensures that the problems
are identified before solutions are advanced.
|
5 FAH-5 Table H-312(2) Data Modeling
|
ACTIVITY
|
PURPOSE
|
|
Data Modeling
|
a. Computer systems manipulate data. Ultimately, everything
a computer does can be reduced to the motion of switchesbinary digits, or
bitsthat are turned on or off. Millions of these switches combine to
effect extremely complicated activitiesword-processing, spreadsheets, or
on-line transactions. For these activities to be effective, the data used
must be organized for efficient use.
b. The standard industry practice for database
organization is normalization. This technique, illustrated below, attempts
to ensure that a single piece of information is stored in oneand only
oneplace, and that information relationships are accurately and
unambiguously represented.
|
For a simple example of data modeling, consider the
following typed list of telephone numbers:
5 FAH-5 Table H-312(3) Non-Normalized Data
Name Type Number
Charlie Brown Home (555) 555-1212
Charlie Brown Cell (505) 444-1212
Linus van Pelt Home (555) 555-2222
Linus van Pelt Cell (505) 444-7474
Linus van Pelt FAX (555) 555-2223
Lucy van Pelt Home (555) 555-2222
NOTE: There are only three names on the list, and three
different types of telephone numbers. Normalizing this list might create a
name table, a phone number category table, and a phone number table as shown in
5 FAH-5 Table H-312(4).
5 FAH-5 Table H-312(4) Normalized Data
Name Table Phone Category Table Phone Number Table
ID # Name ID # Type Name Type Number
1 Charlie Brown 1 Home 1 1 (555)
555-1212
2 Linus van Pelt 2 Cell 1 2 (505)
444-1212
3 Lucy van Pelt 3 FAX 2 1 (555)
555-2222
4 Office 2 2 (505)
444-7474
2 3 (555)
555-2223
3 1 (555)
555-2222
a. Normalization could further occur by separating
first names from surnames, or segmenting phone number area codes (and even by
phone exchanges), or isolating only unique phone numbers. These new attributes
would add additional fields of information about the data elements.
b. The continuing challenge of normalization is to
organize data in ways meaningful to the user while avoiding any repetition of
information. One result is that the data can be retrieved, displayed and
printed in different ways. Another result is that the computer can manage the
relationships between data elements more readily when the data is normalized.
Different people have different telephone numbers, establishing one relationship.
Different people have different categories of phones and thus, the table above
on the right bridges two separate relationships. A third result is the
reduction in data redundancystorage of the same data in more than one place.
Along with this is the reduction in data inconsistency. What frequently
happens when the same data is stored in several places is that the data values
are different, leading to confusion as to which value is correct.
c. In some hardware and/or software environments,
optimized data retrieval might require that data be organized in ways specific
to the environment. This is known as de-normalization. If a project has made
a decision to de-normalize data, the decision and its justification should be
documented for future reference.
d. Data modeling identifies data with data names; it
describes data with data attributes; and it identifies relationships among data
objects, usually referred to as entities. An entity is the item about which
you are gathering data. This graphical depiction of data also identifies data
cardinalitythe quantitative relationship between items; every item A is
related to zero, one, or many occurrences of item B; item B may exist
independently of item A. Again, using the table above, each person apparently
can have as many as four different phone numbers.
e. Graphical data models provide an authoritative map
to the information being managed by a system, answering questions and reducing
ambiguity. Graphical data models are also a great deal easier to understand
than a textual representation of the same information.
f. A data model is a necessary tool for an analyst to
understand the overall requirements for a business process. Each step of a
business process handles data. Data is retrieved, stored, manipulated, and
passed on to another part of the process. For the process to operate
efficiently, the supporting data must be available and structured to
accommodate the process. A data model is a visual way to describe the required
data structure.
g. Further, effective normalization and accurately
recording data management decisions provide flexibility for the system. A
system can be built for one-and-only-one purpose, then very often a need arises
for that information to be effectively moved to another environment. Such
integration is facilitated by effective data modeling.
5 FAH-5 Table H-312(5) Data Mapping and Integration
|
ACTIVITY
|
PURPOSE
|
|
Data Mapping and Integration
|
a. An organization rarely has the opportunity to build
everything at once. It is almost inevitable that data mapping will be
necessary to combine information from two or more systems, in support of
system integration.
b. Data mapping involves clearly understanding the
data in both systems, and then articulating the way in which the data can be
transferred between the systems.
|
To illustrate the issues associated with data mapping,
consider the two data tables in 5 FAH-5 Table H-312(6).
5 FAH-5 Table H-312(6) Data Mapping
System A System B
Area
Name Phone LName FName Code Number
Linus Van Pelt (555) 555-2222 Bailey Beetle 444 444-8686
Lucy Van Pelt (555) 555-2222 Bumstead Blondie 777 707-3030
Charlie Brown (555) 555-1212 Bumstead Dagwood 777 707-3030
h. Data mapping between these two systems will involve
constructing several procedures, known as algorithms, for moving the data from
system to system. If information is going to move from system A into system B,
system As Name field will have to be broken into first and surnames;
likewise, system As Phone field will have to be separated into the Area
Code and Number entities in system B. The opposite procedures would be
required for migrating information out of system B to system A. The
construction of these data tables and algorithms is the data mapping process.
i. It is important to recognize, as well, that a
direct map of system to system is not considered industry best practice because
it permanently links the two systems together in ways that may be
counter-productive. The physical constraints of the two systems become such
that no changes can be made to either system without the changes impacting them
both simultaneously. 5 FAH-5 Table H-312(7) illustrates this direct linkage.
5 FAH-5 Table H-312(7) Hard-wired System Integration
j. By establishing the data standard as the
integration point, each system needs only to continue to maintain the data map
between the system and the standard. The result is conceptually shown in 5
FAH-5 Table H-312(8).
5 FAH-5 Table H-312(8) Mapping Through the Data Standard
k. Using a standard form of name for data objects,
along with a standard form for the data contained in data objects, reduces the
complexity of the algorithm and makes data mapping easier. Thus, the
standardization process described below makes data management easier. For
guidance refer to the Object Definition and Naming Standard, available from the
program office or the Bureau of Information Resource Management website.
l. Commercial off-the-shelf products create unique
problems in data mapping. Because commercial products are designed to address
a specific and finite series of functions, rather than to fit comfortably
within a suite of software systems, additional analysis is necessary to enable
the integration. Industry best practice typically requires that process and
data models of the commercial product be delivered along with the product
itself. Where such documentation is unavailable, it becomes necessary to study
the product at length to generate the background for data mapping to occur.
The total cost of a commercial off-the-shelf product can be raised
significantly by the analysis, modeling and mapping work required to
effectively integrate it into the enterprise.
5 FAH-5 Table H-312(9) Data Quality Analysis
|
ACTIVITY
|
PURPOSE
|
|
Data Quality Analysis
|
a. A business is not merely a collection of processes,
it is also a collection of business rulesbusiness policies that govern its
own behavior and distinguish it from others. These rules govern changes in
the state of the enterprise, and apply specifically to data elements. When
business rules are not clearly articulated, the user community implies
themdifferent users may therefore imply different things, leading to
misunderstandings and error. Data quality, then, is interpreted in
consideration of how consistent data is with the business rules of the
enterprise.
b. Data quality audits can identify the extent to
which a database is consistent with its own business rules, but does not
automatically solve the problems involved e.g., knowing that a business rule
exists that every customer address must contain a ZIP code does not provide
ZIP codes for the 43% of the addresses missing them. In many cases,
enterprises must accept databases audited to internal consistency below 50%
because the time, expense and sheer ability to correct the problems are not
available.
|
5 FAH-5 H-313 DATA STANDARDIZATION
(CT:ITS-9; 01-30-2019)
a. If a program was intended to perform the service
functions described above, it would make a significant contribution to an enterprise.
However, this contribution would be limited if it were not tied together in
meaningful ways. The data administration program, therefore, uses its
service component as the data gathering mechanism for standardization. By
working with actual data use in the Department, data administrators better
understand the data objects, attributes, relationships, cardinality, and
business rules of the Department. By conducting further analysis, data
administration organizes related information and articulates this actual
business usage as the standard for the Department. This provides guidance for
new systems and integration activity. The standardization effort also
maintains flexibility in the enterprise for data reuse and elimination of data
redundancy.
5 FAH-5 Table H-313(1) Enterprise Data Model
|
ACTIVITY
|
PURPOSE
|
|
The Enterprise Data Model (and Standard Data Elements)
|
a. Information gathered about data usage in the
Department moves into the enterprise data model. A continuing work in
progress, the enterprise data model is regularly updated in quarterly
releases of the Standard Data Elements volume [available on the Bureau of Information Resource
Management website. This document provides data models graphically depicting
data objects and their relationships, and articulates standardized data
names, data attributes and business rules relevant to the data objects.
b. The enterprise data model is not intended as a
requirement, but as a statement of how data is used in the Department. An
office wishing to use the model for employee names would probably not use all
of the elements in the person name model, which explains all the data
requirements identified within the Department.
c. The guidance of the data administration program is
that all development and integration activity use the standard data elements
articulated in the enterprise data model whenever possible, and especially
for data integration as shown above. Where questions emerge about how to
apply the enterprise data model in a particular environment, contact the data
administration program.
|
|
|
|
5 FAH-5 Table H-313(2) MetaDataBase
|
ACTIVITY
|
PURPOSE
|
|
MetaDataBase
|
a. The metadatabase is the integrated set of data
tools used by the data administration program to store the information
contained in the enterprise data model. Data models, process models,
relational databases and other forms provide a comprehensive view of data
usage in the Department.
b. System developers may wish to use the metadatabase
repository as a common source of information for system development. This
topic is discussed in the Repository Implementation Guidelines document
published by Data Administration and available on the Bureau of Information
Resource Management website .
|
5 FAH-5 Table H-313(3) Data Administration Standardization
|
ACTIVITY
|
PURPOSE
|
|
Data Administration Working Group (DAWG)
|
a. In order to ensure that data administrations information
about data usage in the Department is generalized beyond one specific office
environment, the Data Administration Working Group meets quarterly to discuss
additions suggested to the standard data elements documentcandidate
standard data elementsas well as other topics of common interest. In these
sessions, recommended data names and data formats are viewed in the context
of other business users, so that the resulting standard can be generally
beneficial.
b. Meetings of the Data Administration Working Group
are open to all who are interested in attending. Database administrators and
data stewards are particularly encouraged to attend.
c. Information or proposals may be submitted to the
Data Administration Working Group by contacting the data administration
program. Questions about the Data Administration Working Group should
likewise be directed to the data administration program at (703) 875-4400.
|
5 FAH-5 H-314 SUPPLY
(TL:ITS-1; 02-13-2002)
Data administration is the resource management function
for Department data usage. As the only program studying data throughout the
Department, data administration is uniquely positioned to identify
opportunities for data re-use. The supply function is the third major
component of the data administration program.
5 FAH-5 Table H-314(1) Standard Data Tables
|
ACTIVITY
|
PURPOSE
|
|
Standard Data Reference Tables
|
In many cases, data has a single source and changes little
over time. In such cases, data administration makes an effort to manage the
data and provide it to the enterprise in useful form. The Data
Administration OpenNet site contains databases of several types of reference
data for general use by users and developers throughout the Department.
|
5 FAH-5 Table H-314(2) Data Stewardship
|
ACTIVITY
|
PURPOSE
|
|
Data Stewardship
|
a. More commonly, the members of a specific business
area manage data. The Bureau of Human Resources manages information about
the employee and the Bureau of Financial Management and Policy manages
financial information. In such cases, it is unnecessary for data
administration to take ownership of the information. Those in the specific
business area become the data stewards, providing access to this information
for Departmental use.
b. Data stewards identify the conditions whereby
business users in certain roles should be allowed to create, read, update,
and/or delete information. They also manage the data quality of the
database. Data stewards facilitate data re-use, and move the Department
closer to the goal of reducing data redundancy while supporting integration.
|
5 FAH-5 H-315 THROUGH H-319 UNASSIGNED