Data modeling is old. It is actually prehistoric in Information Technology terms. People are quick to label pre-internet technologies as dinosaurs and relegate them to the graveyard. Data modeling and the discipline that surrounds it may be old but by no means dead.
What was old is new again. My initial exposure to data modeling was in the late 80s with it becoming my profession in early 90s. Data modeling has continued to evolve over my career. It’s also important to note that data modelers and data modeling practices have evolved over this time. Here are five evolutions I have observed over time that make data modeling new again.
- It was about the picture but now is more about the text.
The data model is the epicenter of the data modeling world. It serves a critical role that has withstood the test of time. There has been a definite shift to focus more on the importance of metadata behind the data model objects. This is largely due to the increasing accessibility of metadata across software and hardware that was largely absent in the early years. With this accessibility come new audiences, audiences that value metadata over a data model.
- It was about a lot of detail but now is about just enough detail.
Methodologies and IT projects have dramatically evolved over time. Gone are multi-year projects where data modelers labored for months on a data model. When agile surfaced, data modelers cried foul. What we have learned over the years is that it makes sense to work in small bites collaboratively with team members from many disciples. This shift means that models may have a varying degree of detail that changes over time and is delivered in pieces.
- It was for a targeted limited audience but today it’s a wider, possibly unknown audience.
From their onset, data models were IT assets in frameworks, methodologies and project plans. The typical audience was analysts, developers, DBAs and other technical folk. Conceptual and logical models involved the business side of the house. Today’s models and metadata have a wider reach. Models manifest themselves in differing views and levels of detail; more so than in days past. Interconnectivity of data results in data model assets being shared with a much wider audience.
- It was modeled on costly proprietary tools but now on more versatile tools.
The evolution of software has been truly amazing; an evolution obvious in data modeling tools. My first experience was with CASE tools: Teamwork, IEW and ADW. Each of these tools was costly bundled in suites costing as much as $100K/seat. The move to server-based tools that allow collaborative development on laptops dramatically improved productivity. This move also opened the door to metadata exchange and reuse between DBMS, reporting and ETL tools to name a few. The data modelers work became visible and usable by a wider audience. Many tools can use the data model design with their tools to improve efficiency.
- It has always been met with resistance but the resistance has evolved.
Through my 25 + years of modeling I have had to defend my job, my deliverables, my purpose, and my authority countless times. Regrettably, I have never been able to brand myself as a superhero of the data world.
In the early years of my career, the resistance came in form of the developers and project team members who were recipients of my work. Looking back, I must admit our methodology permitted me to be that pain in the a$$.
Mid-career, resistance remained with developers but shifted significantly to IT management. It was a time when IT was looking to streamline and cut costs while increasing efficiency. The data model seemed like an unneeded pretty picture.
Today, developers still resist the model. <sigh> The bigger resistance comes in terms of applicability to new technologies. Data modelers are the resisters. They are forced to expand their horizons and look for opportunities in places they have never ventured before.
Old and somewhat new again