The cornerstone of every commercial business is profitability. If market forces
do not permit a great deal of leeway in the pricing of goods and services, other
ways must be found to generate adequate returns. Reducing expenses is one of the
most obvious ways to expand profit margins, and positioning the organization to
react to new business opportunities quickly with flexibility and accuracy is
another.
Continuing to build automated business solutions, whether operational systems
or BI applications, using the current development approach is both expensive and
time consuming. Therefore, a mechanism that can cut costs and shorten the
delivery cycle must be viewed as a strategic tool and a competitive weapon. This
is where a well-supported and fully operational meta data repository environment
can help.
One of the best ways to reduce expenses, and shorten the delivery cycle at
the same time, is to reuse what has already been built. This has been recognized
in the maxim, "Reuse before you buy and buy before you build." For IT to follow
this maxim and to be able to quickly react to changing business demands, IT
cannot continue to remain primarily a builder of new applications or continue to
buy "silver bullet technology solutions." IT will have to evolve into an
assembler of reusable components!
Reusing existing data and processes makes such good sense that it is
difficult to find anyone with serious objections to it; yet in practice hardly
any organization is mandating a reusability policy from their IT group. One
reason for the lack of reusability of components is probably the fact that it is
nearly impossible to accomplish, unless a central facility like a meta data
repository is developed, populated, and used. Therefore, a meta data repository
must be recognized to make a significant contribution to the bottom line of the
business as a whole.
Reusability of data and processes
We need to consider what makes
something reusable. There are two criteria that must be satisfied. In the first
place, what is built must be designed and constructed with reusability in mind.
This applies equally to data as it applies to processes. Secondly, the
components to be reused must be accessible and identifiable as being reusable.
This means that having a list or catalog of relevant information about reusable
data and processes is imperative. A meta data repository is designed
specifically for that purpose. It functions in much the same way as the card
catalog does in a library. The card catalog does not contain any of the books in
the library, but contains relevant information about those books including a
pointer (Dewey Decimal Number) that will allow the user to go to the place in
the library where any particular book is physically stored.
All librarians recognize the value of the card catalog and the need to
maintain it. Since the collection of books is always changing, the catalog must
be continuously updated to reflect these changes. This is an integral part of
the natural processes of the library. It would be difficult to imagine using a
library that did not perform this maintenance, yet most IT organizations are run
without a similar catalog of their objects, namely the data items, data
structures, programs, modules, and applications in their inventory. The
maintenance of the meta data repository must become as integral to the natural
process of managing the IT component inventory of an organization as the
maintenance of the card catalog is to managing the book inventory of a library.
Both are essential if the respective endeavors are to succeed.
Reusability Software Factory Model
The word inventory suggests
business activity. Items in inventory can be purchased or manufactured. Either
way, a business that does not know, with a high degree of precision, what it has
on hand at any given moment, does not have the information it needs to control
its processes effectively and efficiently. Having something on hand and not
knowing it, is worse than not having it and knowing that it is not available,
because buying or building something that is already "in house" wastes time and
money, and in the end produces redundancy as well. This redundancy must be
maintained, thereby adding even more time, money, and resources; all of which
increases expenses and lengthens the delivery cycle.
If, on the other hand, a fully operational meta data repository was
available, a search for an object (data or process) would produce a "hit". The
returned information would confirm that the cataloged object is the required
one. Confidence in the meta data repository as a labor and money saving venture
would be reinforced. Time and money that would have been spent in the
reconstruction of the object would be saved. In addition, a lengthy search for
information that may or may not exist about the needed object is avoided as
well, because the authoritative source about such things, the meta data
repository, was consulted. Imagine how much easier Y2K impact analysis would
have been with a complete IT component inventory in a well maintained meta data
repository.
It is instructive to view the IT organization as a software factory. Just as
the word inventory suggests business, the word factory suggests
dynamic processes and productivity. No senior executive or plant manager would
tolerate the risk and uncertainty associated with the lack of information about
factory processes, that IT organizations seem to accept as normal in the course
of constructing applications. Using the meta data repository in conjunction with
a software factory model or mindset can produce the control information that can
put the processes of application construction and maintenance on a more sound
and business-like basis.
Every technique that a factory uses to control its operation has a
counterpart in IT systems. A factory uses predefined processes for executing
work orders depending on their nature, scope and degree of completeness. These
processes correspond to the development steps, activities, tasks and deliverable
definitions of an application development life cycle. A factory also records
data about work-in-progress as well as information on completed work orders,
inventories of finished goods, shipped products, levels and use rates of raw
materials, etc. This information is indispensable to running a factory
effectively and efficiently. The entire collection of data and information that
a factory uses in managing its business has its IT counterpart in the contents
of the meta data repository.
Additional Benefits of Reusability
In addition to maximizing the value
of available resources, using the meta data repository to reuse data and
processes has additional benefits beyond those already mentioned. Reuse promotes
consistency and less redundancy, which in turn has a major effect on an
organization's ability to be Žuick, flexible, and accurate? When a single source
of data or a single process must be changed, it is "quicker" to make required
changes in one place than in many. Making a change in one place to a data
structure or process also assures consistency and higher quality, which directly
affects being "accurate". By creating structures that are designed from the
beginning for reuse, the "flexible" term in the equation is positively affected
since an approach of "mix and match reusable components" becomes feasible.
When the meta data repository is used to control the business of application
development and maintenance, it now becomes possible for the organization to
make the transition to an information-based environment. Meta data is
instrumental for this transition because of the very simple relationship that
exists between data and information, which is data + meta data =
information. Information is data within context, and meta data provides that
context.
Many companies today find themselves data rich, but information poor. This
condition can be changed by using a meta data repository to define the
information structures a company needs. Capturing the data and information
structures a company has will permit analysts to determine what must be done to
transform existing data into required information. The transformation processes
(algorithms, programs) should be captured as meta data as well. Once this has
been done, the meta data repository will be able to assist in influencing the
future as well as in documenting the past and controlling the present. The meta
data repository cannot by itself make the needed changes, but it is a powerful
enabler if the required contents are available and management decides to use
them to that end.
Executive Sponsorship
A critical success factor for an effective meta
data repository, is to create an infrastructure in which it is used. This
infrastructure is more appropriately called an environment, because
infrastructure implies something that can be physically built whereas an
environment goes beyond that and speaks to ways of operating and thinking as
well. A successful meta data repository environment is one in which current
processes will have been reengineered and attitudes reshaped to support a more
efficient and effective use of resources throughout the application development
life cycle.
Reshaping of attitudes and reengineering the processes currently used in
application construction is the responsibility of the executive management in
the organization. The role that executive management must play is communicate
the importance of using the meta data repository as a way of managing the
organization in general, and the delivery and maintenance of operational systems
and BI applications in particular. This executive sponsorship cannot be
overemphasized.
There are several important parts to this sponsorship. The most important is
the communication of management expectations that meta data will be a
deliverable of every application. Management must explain that it requires an
accurate and complete knowledge base of the IT component inventory, as well as
the business definitions upon which design decisions are based. In other words,
management must establish a policy for mandatory meta data repository creation,
maintenance, and usage.
Another important aspect to meta data repository success is to adopt a
"pay-as-you-go" approach. This means updating the meta data repository as part
of the process of application development and not as an afterthought.
Documenting an application as an afterthought is usually considered "extra work"
by the technicians. But using the meta data repository to document the work as
it is being done in conjunction with a rigorously used application development
methodology is vital to creating robust applications and the information
necessary to maintaining them quickly and at low cost. Instituting active
quality control procedures for development projects that include inspection of
meta data repository entries, as proof that the required work has been done,
should be a high priority.
In Conclusion
Meta data repositories have long been seen as tools
primarily of interest to developers and technicians. Their value to knowledge
workers and business analysts is now being recognized. The strategic value of a
meta data repository to an organization is that of being an enabler for the
transition to an information-based environment, which gives the organization the
ability to be Žuick, flexible, and accurate? Managers will be able to manage
more effectively and technicians can be more productive when the business
meaning of data and processes are more widely understood. Once the inventory of
data and process has been cataloged, the opportunities for reuse will increase,
which in turn will promote consistency and enhance quality. Lead times and the
costs of development and maintenance will shrink as more of the inventory can be
identified as reusable.
For the meta data repository to provide all of the benefits mentioned, an
environment must be created in which its knowledgeable use and careful
maintenance are not just encouraged but required by executive management. It is
management's role to create such an environment in which everyone understands
that the meta data repository must be complete, accurate and continuously
maintained because it is a repository of information about the corporate assets
of business data and the business processes, and the IT constructs which
automate them. To be without a full accounting of such vital assets is to forego
the opportunity of maximizing the value of those assets. It also poses the risk
that the competitive position of an organization could be harmed through
unnecessarily high costs, long implementation lead-times, and low quality levels
in its applications.
About the Authors
Mike Brodie is a veteran in the community of meta data
repository administrators. His first exposure to rudimentary data dictionaries
dates back more than 20 years. Since then he has evaluated, selected, installed,
and enhanced numerous data dictionary and meta data repository products. He also
designed and developed a number of customized meta data repositories. He is
committed to educating the business community on how to maximize the business
impact of meta data on the financial performance of their companies. Mike can be
reached at
Ms. Moss
is president of Method Focus Inc., a company specializing in improving the
quality of business information systems. She has over 20 years of IT experience.
She frequently lectures and presents at conferences in the United States,
Canada, Europe, Australia, and Asia on the topics of Data Warehousing, Business
Intelligence, Customer Relationship Management, Information Quality, and other
information asset management topics, such as data integration and
cross-organizational development. She provides consulting services in data
warehouse and business intelligence assessment, spiral methodologies, project
management, data administration, data modeling, data quality assessment, data
transformation and cleansing, as well as meta data capture and
utilization.
Ms. Moss is co-author of three books: Data Warehouse Project
Management, Addison Wesley 2000; Impossible Data Warehouse
Situations, Addison Wesley 2002; and Business Intelligence Roadmap: The
Complete Project Lifecycle for Decision Support Applications, Addison
Wesley, 2003. Her articles are frequently published in DM Review, TDWI, Journal
of Data Warehousing, Teradata Magazine, and Cutter IT Journal, and her white
papers are available through the Cutter Consortium. She can be reached at