by Jeff Pryslak
Rising need to Leverage Information Strategically
There was a recent paper by Gartner that demonstrated that new business initiatives are creating data models that are in silo’s. These silo’s are meant to be short term or narrow scope use cases, but evolve into more operationally focused systems. Due to this trend, there is a distinct need in IT to start understanding the strategic impact of data within the enterprise that is focused above the project level. This is where the discussion of strategic information modeling begins to take place in support of an initiative such as Master Data Management (MDM) or data quality. What the groups are really reaching for is an extension of information architecture for the IT systems that rely upon business data to be successful. This trend of Enterprise Information Architecture (EIA) is focused on the reuse of metadata about processes, data and rules that are relevant to making the enterprise successful.
Project level data seems to be a common disease in most organizations today. Why is this a disease? because it is curable first of all, but more importantly, it can be prevented before you ever catch it. This is simple to state, but the reality can be more elusive. In the next few pages, a viable approach to implementing EIA and overcoming the project silo will be discussed to the advantage of IT and its need to understand the business more effectively.
Containing the Project Level Disease
If your organization has been diagnosed with this, let everyone around you know that it is contagious and can result in follow on complications, also called growing pains. Once one project has a successful startup in production due to isolation, other projects will see this as the lowest risk approach to follow creating a wave of disconnected data sets with little hope of easy integration. To illustrate this, lets discuss a customer that I have talked to for a number of years. After visiting with them on an unrelated database technology update, we reviewed their architecture for a future application and identified that they had this disease. They needed to bring a number of existing data sets together to extend a business process for creating more value to what was already there. Their goal was to generate a master data management process to incorporate all of the current data dictionaries into one cohesive repository. During this process, they identified over 300 applications that are suffering from this malady. After five years, this effort is still ongoing to my knowledge, but it is improving. How can they tell? Since the executive level decision was made to incorporate MDM, 30 new applications have been brought into production, all of which conform to the enterprise data dictionary. This allows them the ability to easily move data within their organization for enrichment, reporting and archival with much more efficiency.
How did this customer approach their MDM effort? Their initial efforts began with the core process of Data Modeling through reverse engineering of existing databases and XML schemas. This provided a baseline of data entities to start creating definitions. What they were not expecting was the need to create higher level conceptual definitions of these data entities. This need arose when two different data sets had like terms. The whole effort focused on how to agree on what specific terms meant at an enterprise level. Since they were using PowerDesigner, the creation of a Conceptual Data Model (CDM) linked the two distinct Physical Data Models (PDM) together through the use of mapping the terms. This created a lineage between one common data entity in the CDM from two contributing terms in the PDMs. In the latest version of PowerDesigner, the Business Glossary was introduced as an enterprise repository list of terms and synonyms that link to modeling artifacts. This is accomplished through the use of metadata that is created during the generation, selection, mapping and linking of these glossary terms to design objects, such as data entities, database tables or columns, etc…
Prevention is the Key
This process of curing the disease can be extremely costly, but minor steps can be taken to prevent this from happening now and in the future. There are three specific methodologies that any organization can adopt which act as a governance and quality system for the business. First, the creation of a standard list of business terms and their data specification. Second, the integration of a system to check either the data models or information architectures of new applications. Finally, linking the processes, rules and business logic directly to the data that is used to support it. Of these three items, the second is most challenging, but all of them are able to be accomplished in your department today.
The data dictionary, or business glossary, is now a well defined requirement for any IT Department that supports their business through technology. All of those systems create, store and modify data that needs to be leveraged for the success of the business, but they define those data elements in slightly different ways based on the use case. One advance in PowerDesigner is focused on helping IT reduce these inconsistencies by using terms and associating related terms and even synonyms so that the process of data creation can be more informative for existing definitions and information association. In Figure 1, we see a standard glossary which has a number of terms. These terms are then used in the UML diagrams, Physical Data Models and Business Process Models whenever name and code fields are entered. An automated drop down list of glossary terms fills in the field to generate a naming pick list, or offer the ability to create a new term and definition. This greatly enhances the return of the glossary effort within any business due to the metadata being collected on what databases, applications or business processes are using those terms within their implementations.
The next step in creating an effective prevention program is the use of checks in the project management process. Portions of this can be automated with tooling, such as a PowerDesigner Check Model, but this needs to be integrated into a review process for all systems entering the test and production phases within a business. This process integration limits the risk being introduced through IT systems by identifying the level of compliance for applications to existing data availability. The real return on this effort comes a few months later in the life cycle management of the systems when a change or upgrade occurs and the impact of those changes will need to be evaluated.
The resulting metadata from linking business glossary terms to data and other objects within the system landscape allows for a complete introspection of the impact of change as well as the lineage of information within the enterprise. These two items are like the ‘Holy Grail’ to the business due the high value in being able to reduce risk and control costs within IT budgets. How does an architect provide these benefits? This is where the metadata repository is introduced as a value added resource to the Enterprise Information Architecture. In the world of PowerDesigner, this repository has two primary entry points, the tool interface and a web browser. The Portal Composer is the web based access to the repository and allows authorized users to view objects, definition and do analysis on the repository to obtain these impact and lineage answers. In Figure 2, the image shows a tree of impact that a change to the Order entity would have on the enterprise architecture as a whole. Governing this list is a set of propagation rules that can be adjusted based on what you want o evaluate. So, if a project manager is going to change a front office data entry system, they can quickly see all the reporting and data warehouse implementations that will be have to be taken into account for that change in accounting standards. This gives an accurate and risk reducing view into basic change requests that the business might need, but does not comprehend the extent of that impact.
Enterprise Information Architecture links Project Silos
The end result of the above steps is a clear path to accomplishing an Enterprise Information Architecture to add business value to project management through risk reduction and accurate estimates. The intermediate benefits created provide a clearer picture of the gaps between application silos that need to be filled with the metadata the architecture will provide.
Finally, the short term gains seen in this process are standard definitions of terms and data entities that create an enterprise of consistency reducing time to integrate systems that are in production today.
Conclusion: Businesses can become “diseased” by broken metadata connections – an illness not often visible from the outside, but one that costs time, money and erodes competitive edge. The cure, a focus on architecture which aligns metadata, reconnects broken touch points and increases communication and collaboration across the enterprise. With metadata flowing more freely between business process definitions, conceptual data definitions and the logical/physical implementations, an enterprise will begin to restore agility and ‘good health”. By taking the first evolutionary step in this process – you will gain the advantage of standardized business language and consistently of data definitions, which will start to reduce time integrating production systems today, and will grow into an architected enterprise.
About the Author
Jeff Pryslak is a Director of the Evangelist Team at Sybase, an SAP Co. For the past fifteen years in IT, Jeff has focused on security and architecture in numerous customer projects, from US Air Force to commercial banking. This has inspired him to bring PowerDesigner’s support for all architecture environments to global companies and public sector organizations. He also supports the Sybase effort to maintain and enhance its security certifications with groups such as Common Criteria and RSA to meet the ever increasing risks of computing on the internet and in the cloud.