Moss L.T., Atre S. Business intelligence roadmap: The complete project lifecycle for decision-support applications
Подождите немного. Документ загружается.
the new releases and versions of the licensed meta data repository product.
Staffing Challenges
Meta data should be "living" documentation stored in a database, that is, in the meta
data repository. Storing meta data as paper documents is guaranteed to turn it into
"shelfware" within months, if not weeks. This means that, at a minimum, one meta
data administrator must be dedicated full-time to managing the meta data repository
content and the software. If a meta data repository is being built as part of the BI
project, a staff of one person will not be enough. The meta data repository effort will
require an analyst, a data modeler, a database designer, and one or more
developers.
Budget Challenges
Although many BI experts think of meta data as the "glue" of the BI decision-support
environment, most organizations allocate little or no money for creating and
maintaining a meta data repository. They still regard meta data as systems
documentation for technicians, rather than a navigation tool for business people. The
pain of access frustration and data confusion must often reach an intolerable level
before organizations include meta data as a mandatory and standard deliverable of
their BI projects.
Lack of meta data has frequently been cited as one of the
reasons for BI application failure.
Usability Challenges
Using a meta data repository should be completely intuitive. Business people should
be able to click on an icon and immediately get the requested information about a
table or column, a chart or report, or even a business query. More complex inquiries
against the meta data repository should be handled with built-in or customized
macros. However, the most polished way to present meta data is to include it in BI
queries, as shown in Figure 7.6.
Figure 7.6. Example of Meta Data in a BI Query
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
241/631
Unfortunately, many meta data repository products are still designed by technicians
for technicians rather than for business people. Some of these products still have a
cryptic meta data language, lack sophisticated reporting capabilities, are not context
sensitive, and require an understanding of the meta model that describes the meta
data objects and their relationships.
Political Challenges
Building an enterprise-wide meta data solution is difficult because departmental
differences must be reconciled and cross-departmental politics must be resolved.
These disputes, although totally predictable, are rarely taken into account when the
project plan is created. As a result, projects are delayed while these issues are
addressed or pushed up to business executives and steering committees. This gives
the impression that BI projects are difficult, controversial, tiresome, draining, slow,
and generally undesirable work.
Despite all these challenges, a meta data repository is a
mandatory component of every BI decision-support
environment.
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
242/631
[ Team LiB ]
The Logical Meta Model
Regardless of whether the meta data repository is licensed or built, and regardless of
the implementation method (centralized, decentralized, or distributed, as discussed
in Step 10, Meta Data Repository Design), the meta data repository should support a
logical meta model, which reflects the meta data requirements. As with business
data, each component of meta data is unique by nature. It is important to define
these unique meta data objects, their contents, their interrelationships, and their
interdependencies, independent of how they will be stored or accessed. The
technique for this activity is logical data modeling, only in this case it will produce a
logical meta model.
A logical meta model is a data model that indicates objects, the relationships
between the objects, and the cardinality and optionality of the relationships. The
difference between a logical meta model for a meta data repository and a logical
data model for a business application lies in the nature of the objects. Objects in a
logical meta model represent meta data, such as entity, attribute, definition, domain,
table, column, and index. Objects in a logical data model represent business data,
such as customer, product, employee, account, and location.
The Entity-Relationship Meta Model
A logical meta model is created during the first meta data initiative and is expanded
with each subsequent initiative. The logical representation of meta data objects
should be captured as an E-R diagram because of its explicit definitions of the meta
data objects, the relationships among them, and the contents of the objects. Figure
7.7 shows an example of an E-R meta model.
Figure 7.7. Entity-Relationship Meta Model. (Short vertical
lines indicate "one," and the crow's feet indicate "many.")
An E-R meta model primarily helps people understand, communicate, and validate
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
243/631
the meta data requirements. Therefore, an E-R meta model should be viewed as a
requirements model to be used for evaluating meta data repository products and for
setting a baseline when designing a customized meta data repository, even if its
physical meta model (database design) ends up being object-oriented (OO).
Meta-Meta Data
Since meta data is the contextual information about business data, meta-meta data
is the contextual information about meta data. Many components of meta-meta data
are similar to those of meta data. For example, every meta data object should have
components that cover name, definition, size and length, content, ownership,
relationship, business rules, security, cleanliness, physical location, applicability,
timeliness, volume, and notes. The meta-meta data for a meta data object might
look like this:
Name: Entity
Relationship: related to one or many tables
Security: read by all, updated by the data administrator
Ownership: the data administrator
Origin: ERWIN CASE tool
Physical location: MDRSYSENT table
Cleanliness: 2 percent missing data
Timeliness: last updated on November 1, 2002
Volume and growth: 2,391 rows, growth rate 1 percent annually
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
244/631
[ Team LiB ]
Meta Data Repository Analysis Activities
The activities for meta data repository analysis do not need to be performed linearly.
Figure 7.8 indicates which activities can be performed concurrently. The list below
briefly describes the activities associated with Step 7, Meta Data Repository Analysis.
Analyze the meta data repository requirements.
Work with the business representative to determine and prioritize the meta
data requirements for your specific BI project. Indicate which of the meta data
components are mandatory, important, and optional. If a meta data repository
already exists, determine which meta data components need to be added, if
any. Update the latest version of the application requirements document
(revised during or after prototyping).
1.
Analyze the interface requirements for the meta data repository.
Whether a meta data repository is licensed or built, it must accept meta data
from different sources. Business meta data will have to be extracted from CASE
tools, word processing documents, or spreadsheets. Technical meta data will
have to be extracted and merged from database management system (DBMS)
dictionaries, ETL tools, data-cleansing tools, OLAP tools, report writers, and
data mining tools.
2.
Analyze the meta data repository access and reporting requirements.
Populating a database is meaningless unless the content can be accessed,
queried, and reported. This is as true for meta data as it is for business data.
Identify the meta data access requirements, security requirements, and help
function requirements. Evaluate alternative display modes, such as Portable
Document Format (PDF), Hypertext Markup Language (HTML), SQL, canned
queries, or proprietary meta data repository reporting software. A context-
sensitive help tutorial would be a beneficial feature to include.
3.
Create the logical meta model.
Draw the logical meta model as an E-R model to explicitly show the
relationships between meta data objects, even if you plan to implement the
meta data repository as an OO database. In other words, the logical meta
model should always be an E-R model, while the physical meta model (the
meta data repository database design created in Step 10, Meta Data Repository
Design) can be either an E-R model or an OO model.
4.
Create the meta-meta data.
While the logical meta model shows the meta data repository requirements at a
glance, the meta-meta data describes the required meta data components in
detail.
5.
Figure 7.8. Meta Data Repository Analysis Activities
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
245/631
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
246/631
[ Team LiB ]
Deliverables Resulting from These Activities
Logical meta model
This data model is a fully normalized E-R diagram showing kernel entities,
associative entities, characteristic entities, relationships, cardinality, optionality,
unique identifiers, and all attributes for meta data repository objects.
1.
Meta-meta data
The meta data entities and attributes from the logical meta model must be
described with meta data. Meta data–specific meta data components (meta-
meta data) are meta data names, meta data definitions, meta data
relationships, unique identifiers, types, lengths, domains, business rules,
policies, and meta data ownership.
2.
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
247/631
[ Team LiB ]
Roles Involved in These Activities
Data administrator
The data administrator gathers the business meta data in a CASE tool during
the logical data modeling activities. This meta data will be one of the sources
for the meta data repository. The data administrator, in collaboration with the
meta data administrator, writes and publishes the data standards. He or she
may also assist with creating the meta model and the meta-meta data.
Meta data administrator
The meta data administrator has primary responsibilities for storing and
providing access to the meta data and for maintaining the meta data
repository. He or she must analyze the meta data requirements, identify the
meta data components, and produce or enhance the logical meta model and the
meta-meta data.
Subject matter expert
The subject matter expert participates in this step by representing the business
people and their meta data requirements. The subject matter expert identifies
security requirements and data ownership and works with the data
administrator, the meta data administrator, the data owners, and other
business people to standardize names, definitions, content, and business rules.
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
248/631
[ Team LiB ]
Risks of Not Performing Step 7
Since one of the BI decision-support objectives is to eliminate inconsistencies, the
source data must be standardized. Standardization invariably results in changing
much of the source data. Changes may include renaming the data, splitting one
source data element into multiple target columns, or populating one target column
from multiple source data elements. It can also mean translating codes into
mnemonics, standardizing (changing) data values, and filtering out inappropriate or
invalid data. In the end, business people will not be able to reconcile their
operational source data to the BI target data unless they have a trace of these
changes. This trace is called meta data, and business people need it to navigate
effectively through the BI decision-support environment.
Without meta data, the business people would have a difficult time understanding
and using the transformed data in the BI target databases. It would be as frustrating
as aimlessly driving a car for weeks or months without a map, guessing your way to
your destination. Once the business people perceive the BI application as difficult to
use or they think the BI data is unreliable because it no longer matches the source
data in the operational systems, they could label the BI decision-support initiative a
failure.
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
249/631
[ Team LiB ]
Bibliography and Additional Reading
Adelman, Sid, and Larissa Terpeluk Moss. Data Warehouse Project Management.
Boston, MA: Addison-Wesley, 2000.
Brackett, Michael H. Data Resource Quality: Turning Bad Habits into Good Practices.
Boston, MA: Addison-Wesley, 2000.
———. The Data Warehouse Challenge: Taming Data Chaos. New York: John Wiley &
Sons, 1996.
Devlin, Barry. Data Warehouse: From Architecture to Implementation. Reading, MA:
Addison-Wesley, 1997.
Hackney, Douglas. Understanding and Implementing Successful Data Marts. Reading,
MA: Addison-Wesley, 1997.
Inmon, William H., J. D. Welch, and Katherine L. Glassey. Managing the Data
Warehouse: Practical Techniques for Monitoring Operations and Performance,
Administering Data and Tools and Managing Change and Growth. New York: John
Wiley & Sons, 1996.
Marco, David. Building and Managing the Meta Data Repository: A Full Lifecycle
Guide. New York: John Wiley & Sons, 2000.
Reingruber, Michael C., and William W. Gregory. The Data Modeling Handbook: A
Best-Practice Approach to Building Quality Data Models. New York: John Wiley &
Sons, 1994.
Ross, Ronald G. The Business Rule Concepts. Houston, TX: Business Rule Solutions,
Inc., 1998.
Simsion, Graeme. Data Modeling Essentials: Analysis, Design, and Innovation.
Boston, MA: International Thomson Computer Press, 1994.
Sperley, Eric. The Enterprise Data Warehouse: Planning, Building, and
Implementation. Upper Saddle River, NJ: Prentice Hall, 1999.
Tannenbaum, Adrienne. Metadata Solutions: Using Metamodels, Repositories, XML,
and Enterprise Portals to Generate Information on Demand. Boston, MA: Addison-
Wesley, 2002.
Data Management Association (DAMA): http://www.dama.org
DM Review magazine: http://www.dmreview.com
Enterprise Warehousing Solutions, Inc.: http://www.EWSolutions.com
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
250/631