Rebalancing the Data Modeling Methodology

Building the data model used to be relatively easy when the models were smaller and didn’t tend to scale beyond a single subject area or the support of a single application.  There was not much interaction with the business-side interests since modeling was considered a ‘black art’ that the business could not relate to.  It was considered a technical function that a single modeler was charged with building. 

However, without scaling data models, shops ended up with many databases and many versions of each area of the model.  The industry moved to data warehousing to gain some economies of scale.  Without the business involvement, the models, and their corresponding applications, were seldom sufficient to meet business needs until a later stage revision, when finally business interests were somewhat understood.  Even then, the model represented a compromise by the business.

Fast forward to today. Most modelers are working on models that are too large and with too many demands on it for one modeler to control the entire model.  Most models will also have a lifespan much greater than any single modeler’s tenure on the project.  Consequently, there are shortcomings in each model out there and the modeling discipline today, more than ever, needs to have a strong methodology in place for minimizing those shortcomings.

Stepping up to the challenges also means that modelers need to have a compromising attitude – not compromising the methodology, but occasionally perhaps compromising their own modeling proclivities that fall outside of the methodology.  Without a modeling methodology today that maintains the standards for each significant data model being worked on in the enterprise, the enterprise will suffer as modelers reengineer existing models to their standards, rather than adhere to the standards set for the model.

Putting in place a methodology that works for today’s business needs mean incorporating some agile components.  With a balanced, agile modeling methodology, you account for a lasting architecture and fast business results. Too often, methodologies are unbalanced.  I will take you to the two extremes I’ve seen.  The overly rigorous architecture is characterized by copious:

• Project Management
• Change Management
• Time Management
• Documentation
• Testing
• Budgeting
• Planning

Obviously, these are all valuable, but they are means to an end – a data model that supports applications, access and operational business intelligence that drives business ROI.  This unbalanced approach leads to several unhealthy factors in place, which can seriously inhibit data modeling:

• Analysis Paralysis
• Excessive Bureaucratic Overhead and Cost
• Inflexibility
• Stagnation
• A Frustrated User Community
• Excessive Interdepartmental Barriers and Boundaries
• “CYA” becomes the prominent attitude
• Poor employee retention and motivation
• Lost creativity, ingenuity and enthusiasm
• Extended project timelines

If any of these sound familiar, the methodology may need to be tuned towards more creativity and flexibility.  This leads me to the other unbalanced methodology, the overly open methodology.  Usually, shops characterized by these factors have given little thought to methodology.  These shops are characterized by:

• Creativity
• Openness
• Quick Responsiveness
• A focus on ad-hoc support
• Flexibility
• Reactivity
• Ability to assimilate new technology

Again obviously, these are all good to have.  However, when the scales are tipped too far in this direction, there are some downsides:

• Difficult to Validate ROI 
• Excessive Breakage and Rework 
• Limited Interdepartmental Communications
• Extended Work Hours for Employees
• Low Confidence Level with the User Community
• High Stress Levels Among Employees
• Over Commitment to the Business Community
• High Support Cost 
• Many Unknowns in Problem Resolution
• Little Time for Employee Development

The trick is to balance the methodology so that you are afforded the best of both worlds.  With such a balanced methodology, you maintain the structure that is required to prevent hitting a wall while, at the same time, being agile and able to keep up with the pace of business need.  You adhere to both the short-term customer service aspect of modeling while also adhering to the higher calling and responsibility of information management.  Though these forces can tend to seem at odds, in reality you do your customers no favors by only responding to their immediate needs and not maintaining an environment that will be able to respond to their needs next week, next month and next year.

A balanced methodology creates modeling environments that are characterized by:

• High Quality, Low Maintenance Models with Minimal Breakage
• Limited Model Rework
• Fewer Working Hours with more Productivity
• Low Stress more Predictive Environment
• Resource Leveling (just enough resources in the right places)
• On-time and on-budget results, resulting in a more satisfied business community
• Easier Justification Processes and Business Approval of New Solutions
• More Opportunity for Career Path, Employee Development and Training
• Faster Problem Resolution with Calculated Predictive Research Actions 
• Realistic Expectations Set at all Levels with Open Communications
• IT Viewed as a Strategic Business Unit at the Corporate Level

These are some of the results when you have an effective modeling methodology in place.  Many times the actions to move the methodology become obvious once the current situation is seen in light of its shortcomings and opportunities missed.