Most data management practitioners are aware of the challenges in orchestrating an effective and collaborative conversation between business and IT professionals about corporate data. What’s required is an on-ramp to engage business users in a meaningful conversation about data, which will enable them to “paint a picture” of the data that is important to them.
To be useful, this picture needs to represent the system’s technical metadata – accurately capturing the business interpretation of the data buried deep within the system architecture. Data modeling is often used as the “palette” of choice to convey this business view of data.
Generally speaking, data modeling has four layers (see figure), each of which serves a distinct purpose and audience. The first two layers – subject area and conceptual models – are intended for conversations and communication with business users. These layers convey the main business and data concepts, highlight key entities and attributes, and are ideal for communication (but not for database design).
The bottom two layers – logical and physical – are more for IT design, development, communication, and execution purposes.
In the past, leveraging data models as a communication vehicle has achieved mixed results. Common criticisms are that the modeling output is too far removed from the physical system logic, making it difficult to act on anything on the modeling output. And once the models are done, they are too static and hard to update or change based on new business requirements or changes to the system of record. The good news is there is a way to address both challenges.
Reverse-engineered, hybrid data modeling
With the right mix of people, process, and technology, you can develop and deploy an effective top-down/bottom-up hybrid modeling solution. For example, you can leverage two best-of-breed applications to effectively bring business and IT users together to effectively collaborate and capture the essence of the business’s priorities. And you can run multiple parallel modeling workshops for different systems where all teams can arrive with the same consistent look and feel for every model, for every system.
The result of this business and IT modeling collaboration provides a clear line of sight and traceability from the priorities identified by the business, to the tables and fields in the originating systems of record.
Best-of-breed technology platforms
By combining the unique features and functions of an information governance solution and a case modeling tool (UML design tool), you can quickly create a comprehensive, integrated solution.
To do so, your team needs to develop a custom toolbar and import/export templates integrating both solutions. Combined, these assets ensure that 1) data models are created with the same consistent look and feel and, 2) data exchanged between the two applications is done in a comprehensive, controlled, and validated way.
Data modeling solution workshop methodology
You should consider engaging data specialists familiar with both the granular data captured in systems of record, as well as information governance and case tool integration. The data specialists need to lead facilitated workshops with organization business, information management, and IT stakeholders to create subject area and conceptual models for any system.
Additionally, the data specialists need to follow a hybrid – top-down and bottom-up – approach where models are reverse-engineered from system data dictionaries and technical metadata, and multiple sources of input are leveraged for validation throughout the modeling process.
Data modeling solution approach benefits
With a tight technology integration and methodology established, your company can benefit greatly from data modeling, including engaging system business and technology stakeholders alike to effectively collaborate in a hybrid modeling approach. Another key benefit is rapidly creating data models via a dynamic modeling solution infrastructure. This makes it easy to change and refresh models on a continuous basis (with applied, appropriate governance). And it enables objectively linking what is modeled at the business level – subject area and conceptual level – to where the related data resides in physical systems of record.
Finally, you can effectively leverage the modeling output as input for related data-management initiatives, including (but not limited to):
- Subject area and entity definitions => business glossary
- Classified entities => centralized master data creation and maintenance
- Unique entity ids => algorithmic cross-system entity matching
- Data elements => data quality requirements
- Entity associations => business rule requirements for current and future state systems
- Subject area models => structured data governance organization/business data steward roles.
By using the hybrid data modeling approach to create data model diagrams, teams can effectively create a reliable “line of sight” for their corporate data. Using the data from physical systems of record helps to rapidly establish and objectively link items of importance identified by business users to where these things actually reside in physical systems of record.
The end result is a picture of the forest – as well as the trees – that can provide tangible input to a road map of enterprise data management initiatives.