The reports of the death of data modeling and data architects are greatly exaggerated. It seems fitting to end the year sharing my thoughts on the viability of data modeling. Data is changing as it has over my career and so is data modeling. Here are five good reasons to not put on your Sunday best for data modeling’s funeral.
- Relational databases are not going away anytime soon.
It is true that big data is eating away at the RDBMS market share. It makes sense to match a business need to the database that gives the biggest payback. That happens to be big data in many cases. The same applies to RDBMS databases. They still do the heavy work of payroll, accounting, financial applications and many day-to-day operational systems. Data modeling is key in the design and deployment of relational databases.
- People still need to understand data.
Corporate America has not stopped building workhorse applications of day-to-day operational systems to keep the business running. Data requirements still need to be captured. The data model is the most efficient tool to do this. Data modeling tools are mature and produce outputs that speak to data audiences at all levels. The data model is a map, and that map tells the story of data in a few screens or pages. Data modeling is part of most major employers’ methodologies and will remain so as long as it adds value to the development lifecycle and the enterprise data asset.
- Data modeling tools are embracing new data technologies.
Data modeling tools have matured their components beyond the ERD. They have reached deeper into the metadata management and metadata repository space where they deploy metadata through portals and web publication. Major data modeling players are not sleeping through the current wave of technology change. They are exploring ways to protect and expand their market share of their tools for big data projects. The big data space is growing and maturing. Part of the maturity will be the expansion of tools to manage and structure big data. It sounds like an opportunity for vendors and technologists already working in the modeling space.
- Data architects may not look the same
Data architects tend to be set in their ways and what they do. Today’s data architect needs to go beyond the ERD and be more of a resource that explains the enterprise data asset and architecture to a variety of business users. Part of unchaining ourselves from the data model is the recognition that metadata is needed more and more to understand seldom accessed or fuzzy data. As a data architect, you may be called to model data in places where you have never ventured and provide that information in something other than an ERD. Data modeling is not solely moving from a conceptual to a logical to a physical data model in a development project.
- There is a shortage of data architects.
Many data architects began their modeling careers in the 80s and 90s. These folks are retiring and leaving the job market in the coming years leaving a gap in the data design lifecycle. Companies that run their bread and butter operational systems on RDBMS databases scramble to hire a data architect to keep database design, development and maintenance running. Gone are the days when data architects and DBAs were glamour jobs. I am actually not quite sure that was ever the case, but the number of people entering IT pursuing a data career is shrinking. Google, EBay and software startups have enticed college graduates to pursue application development careers that could yield them bigger pay and more glory. This all adds up to giving more life to those who seek the data path in their careers.