The Imperfections of Data Modeling

by Joey L
The article mainly focuses on how limited data modeling can be for designing databases.  Most often, modelers assume that the knowledge around data modeling is all that is needed to prosper a successful database.  In actuality, a successful database design does not encompass completely from data modeling.  The author suggests that one of the areas where data modeling is insufficient, involves the use of code tables.  Examples of code tables include: “County Code, Product Category, Customer Type and so on.” Many of these tables are unnecessary to implement a database.  Some of the code tables have nothing to do with the database design and should be left out until the system or database goes live and running.   Similarly, indicators such as flags and switches are commonly used in database designs with values like “yes” or “no”, though the representational scheme only embodies design rather than purely representing data.  Another issue that is related to data modeling tools involves specifying if columns are null or not null in data models.  Many ER Tools do not address the requirement; they ask users if an attribute is nullable or not nullable but cannot record if a column must be null under special conditions.  This limits how precisely data from the data model can be in a database design.

This article relates closely to this week’s topic: entity relationship modeling.  I find this article to be quite interesting.  The author gives out claims about why he thinks that the current ER tools (as of 2006) are really limited to what data they can encompass.  Many of the “criticisms” about data modeling are related to the structure of how data models are implemented.

A suggestion to help clear these problems from the article is to follow an agile data-modeling scheme.  This method developing a data model is very efficient.  Agile modeling takes on an increment development activities on modeling.  Models are first simply designed efficient enough to allow incremental development in the future.  Since the agile data model scheme only takes on the minimal requirements for a project, the data model could be finished at a faster rate.  As we advance in technology, new data modeling schemes will be designed and hopefully will clear the problems that were addressed in this article.

Chisholm, Malcolm (October 2006). Is Data Modeling Sufficient for Database Design? Retrieved from http://www.information-management.com/specialreports/20061024/1066475-1.html?pg=1

 

 

 

 

1 thought on “The Imperfections of Data Modeling”

  1. Data modelling can be cumbersome. The problem is trying to implement changes when so many people are used to what we have now. Hard to get people to change their ways, you know?

Comments are closed.