While many data modelers are very open minded and not righteous at all in their attitudes, over the years, I have heard the following statements (paraphrased) from data modelers numerous times:
‘There is one and only one best way to model a data modeling requirement.’
‘My data model design is the right way to go. Those other data model constructs and solutions are just not right.’
‘Your application project needs to implement our enterprise-wide standardized model exactly how we modeled it, without exception.’
‘Your data governance program needs to adopt and use 100% of our enterprise-wide data model semantics for consistency.’
‘Our template data model design is the ‘best practice’ way to model this structure and is the right model.’
In my experiences, this attitude and paradigm of being ‘dead’ right has led to the death of numerous data modeling and data integration efforts. This way of thinking that there is one right and best solution is a rigid way of thinking and can lead to separation and walls between people. I believe that the biggest causes for data silos are people silos.
For example, I have been on several data modeling efforts where data modelers have argued vehemently that their way is the ‘best’ way to model the data requirements. On one effort, this led to huge additional expenses and time overruns. Rather than acknowledge the pros and cons of various ways to model a specific requirement[i], the modelers fought over being ‘dead’ right, and this led to a ‘dead’ and cancelled data modeling program.
Another scenario that I have witnessed many times is an enterprise data modeling team that approaches a project to strictly enforce their data modeling standards. While moving towards standardized data structures is a good practice to facilitate consistency, sometimes it is taken too far. For example, I have seen projects where it makes sense to use some of the semantics of the enterprise data model, but in some cases, it is appropriate to deviate from the model because of the specific circumstances. Hence, a balanced perspective is often wise, keeping in mind the overall enterprise data standards as well as the project’s individual needs.
Finally, re-usable templates or, as we call them, ‘Universal Data Models ’ can be misunderstood and/or misused. Organizations often want a predesigned data modeling solution that is already completed for them, sometimes with all the coding done as well, such as a data warehouse or master data management prepackaged solution. Template models can save a great deal of time and cost; however, each organization does have specific needs, semantics, and circumstances that are unique, and it makes sense that a data model should be customized for the requirements and situation at hand. Thus, an appropriate way to use template models is to review these as examples (ideally several different template models to get various ideas) and decide how various data model constructs may be applicable, how they can be customized, or if the templates are just not applicable. I often say jokingly that the Universal Data Models[ii] that my organization licenses are ‘not the right models’! This is because I believe that there is no such thing as the (only) ‘right’ model and there is not one and only one right way of modeling something.
Template models are valuable for showing possible ideas and ways that you may not have considered for modeling data model constructs. They can provide ways that have worked for other organizations as well as jump-starting or quality assuring/checking data models; however, they do not replace the skills of a good data modeler. I believe that it is important to appropriately evaluate alternatives and weigh the pros and cons of different ways to model a requirement in order to make effective data modeling choices. Having templates or ‘Universal Data Models ’ can help provide these alternatives and possibilities that have worked for other organizations.
Thus, I believe that we, as data modelers, need to be careful of being ‘dead’ right, and instead, realize that there are often several possible choices and truly appreciate these various perspectives.
[i] For example, there are various alternatives and possibilities published for modeling the same requirements of many common data model patterns in ‘The Data Model Resource Book, Volume 3’, by Len Silverston and Paul Agnew, published by John Wiley and Sons. The book also provides a framework and templates for making data modeling choices between different levels of generalization.
Len Silverston is a best-selling author, consultant, and a fun and top-rated speaker in the field of data modeling, data governance, as well as human behavior in the data management industry, where he has pioneered new approaches to effectively tackle enterprise data management. With over 30 years of experience as a data management consultant helping organizations world-wide, he is well known for his work on "Universal Data Models", which are described in his The Data Model Resource Book series.
Len Silverston is a best-selling author, consultant, and a fun and top-rated speaker in the field of data modeling, data governance, as well as human behavior in the data management industry, where he has pioneered new approaches to effectively tackle enterprise data management. With over 30 years of experience as a data management consultant helping organizations world-wide, he is well known for his work on "Universal Data Models", which are described in his The Data Model Resource Book series.