The practice of improving data governance in data models is inefficient and can cause a potentially dangerous situation.
This practice poses certain challenges when there are new hires within the modeling area who must be initiated into the company’s specific data modeling methods, vocabulary, and subject area content.
What is the Problem & Solution?
Hint! They are Both in the Models
The Main Issues:
— Lack of Trust in the Metadata
— Excessive Timeframe for reviews, corrections, and subsequent rework
— Inadequate Communication, Monitoring, and Measuring of data model content
The Solution:
Incorporating data governance measures as an integral part of the modeling effort is the key to solving both dangerous discrepancy issues as well as reducing the excessive time it takes to model and introduce new hires into the modeling environment. When governance processes are employed during the modeling process, metadata will be verified and managed for change. When clear, accessible standards, processes, and rule documentation is easily available, each resource is empowered – without the persistent need to occupy a senior resource’s schedule.
Mission Impossible: Data models represent the company’s metadata – and metadata is at the heart of our business’ purpose, integrity, and profitability. Let’s bring the technical aspects of data governance into the modeling function as a requirement, while making the process as painless as possible.
Should You Accept This Mission …
You and your loyal team will develop and/or buy naming standards and data management processes for use across the enterprise (Somebody has to do it!). And the processes must be as simple, yet as effective as possible. When common sense and a ‘minimalist’ approach is the underlying method for developing a modelling ‘infrastructure,’ enforcement will not be an issue.
Data Governance in Models Doesn’t Just Happen – The Requirements
The To-Do List
- Devise a strategy to ensure metadata consistency – without which the organization will continue to create and propagate metadata that lacks quality. Incorrect analysis leads to costly business operations, bad decisions, and sometimes legal risk due to erroneously developed production metadata. The strategy needs to promote trusted, reusable objects to ensure reliability in the standardization of objects and reduction of analysis time.
- Buy/Build a Metadata & Model Management Infrastructure to support Metadata Strategy Goals
Data governance can and must be integrated into the modeling process via standards and processes that guide data design – bringing clarity, consistency, and discipline/structure to the enterprise modeling environment. No matter the organizational structure (potentially multiple departments producing multiple project data models), all models represent enterprise data – the inclusion of ‘enterprise’ standards unifies disparate structures over time and assists in the integration of metadata between systems.
Data governance promotes and ensures excellent data design and prevents the introduction of data defects at the model source. A model management infrastructure, or framework, is the means to deploying and maintaining a data governance/modeling program.
The infrastructure must be:
- Easily accessible (intranet based)
- Easy to navigate, understand and execute
- Efficient – each critical process should ask only for the minimal amount of information necessary
- Enforced – adherence to standards and processes is possible when they make sense, are well-written and are streamlined down to critical elements. Approved, reusable structures reduce analysis and development time – which benefits everyone – we reduce the cost of each project while ensuring the ability to integrate standardized metadata
Integrating Data Governance into the modeling process must be managed as a collection of tasks including reviews and approvals. The management is too complex to be done on an ad hoc basis. A ‘Modeling Infrastructure’ must be developed to provide a structure within which governance can thrive. The content in the infrastructure framework should be well-structured, well-written, and client specific.
The Infrastructure must include:
- Accessible Communication Mechanism (the company intranet)
- Data Object Naming Standards, Model Management Procedures, Forms & Model Templates
- A strategy for component reusability – reducing the time for metadata analysis/development while gaining consistency for both Agile and Waterfall development
- Easy maintenance process – Things change. Processes and standards may/will be modified or added – a metadata and model management infrastructure is a living environment. Change is inevitable… our documentation must be easily modified and published
Additionally:
- The framework’s content must be written from an enterprise view – bringing corporate standardization to the modeling efforts of various departments and project teams
- The use of a data modeling tool that has been (or will be) standardized across all modeling groups is required
- Knowledge Availability becomes a reality when the infrastructure resides on the web – let’s empower the new hires ‘to get up to speed by themselves, and allow the mature resources to do their work without interruption
- Assign the most qualified resources to each task – we’re all naturally better at some things. Use people’s strengths: if someone is detail-oriented, make them the Data Model Repository Administrator (in whatever tool) – people perform magnificently when doing what they like
Bad Data is No Longer Acceptable
Data Governance is no longer optional – the inability to integrate metadata into various systems, measurable amounts of time and money lost on incorrect analysis, and the very real consequences of bad data is no longer acceptable. Ultimately, the bottom line is affected negatively as we risk extensive rework, confusion, and customers who are not being well served.
– – – – –
Written with Robert Lutton. Robert runs Sandhill Consultants North America and had been directly responsible for creating an organization that delivers technology, service, products and training in the areas of Enterprise Data Governance and Data Architecture.
For anyone interested in learning more about how to solve these issues, please feel free to click on Sandhill’s quick ‘Data Model Assessment’ to see where your organization stands relative to industry standard best practices, click here.
Marcie and Axis, as a partner of Sandhill Consultants, offers ‘EM-SOS!’™ 2.0 which provides a Framework for the governance of Data Model Life Cycle strategies, model templates, critical standards and procedures to guarantee a governed, customizable model management infrastructure.
