This week I was a panelist on Dataversity’s Big Challenges in Data Modeling webinar moderated by Karen Lopez. It was a lively discussion of data modeler challenges with a good bit of interaction from the virtual audience. Several questions were posed during the hour. A few questions remained unanswered due to time constraints.
Here are my thoughts on the question, “How do you establish reuse of your data models?” This question originated from our discussion on how much design a data modeler actually does. We talked about reuse, industry models and modeling patterns.
Tom’s 10 Random Thoughts on Data Model Reuse
- Find your pattern.You need to have a repeatable pattern. You can start with data modeling patterns from a book or website. Better yet, develop your own pattern and reuse, reuse, reuse.
- Think like an industry modeler.Model your best practices. Get feedback from coworkers and peers in your industry. Incorporate them into your models. A sound design is more likely to be reused and used as a pattern.
- Don’t skimp on the metadata. Your model will only get reused if your peers and clients understand what the model does and what each component means. Accurate and usable metadata makes you a more reliable source for a reusable design.
- Think in an enterprise way.Don’t let your library of models be shelfware. Reuse components and patterns. At the same time look for improvement and refinement of the existing design. Data models should live on and evolve when needed.
- Model specifically but with flexibility.Data modelers are tasked to model the business. Never forget that key part of your work. At the same time, look for commonality across objects and model to capitalize on common objects.
- Publish and expose your data models.Out of sight and out of mind is a sure way to never reuse a data model again. Develop a means to publish your diagrams, glossaries and data dictionaries. Organize and provide for a way to search and find your data model contents.
- Educate on reuse and commonality. Discuss your design decisions with your business users and peers. Explain the benefits to them and your enterprise. Your commitment for reuse will echo with them and their work. Everyone wants to simplify their work and help the bottom line.
- Establish domains and reuse them. Down to the nitty gritty here; define common domains to use in your enterprise. Defining attributes with these base characteristics bring order to your design. This seemingly small exercise yields big results in commonality across your models.
- If you talk the talk, you have to walk the walk.It is more than just a line in your performance review to reuse models where appropriate. It is about looking for reuse in every modeling effort and in every design decision. That library of models and modeling patterns you developed should be the first place you look when modeling.
- Set and live your standards. I have seen beautifully crafted standards and guidelines on data model reuse that are largely ignored. Post those standards on your cubicle wall. Use them daily. More importantly, revisit and makes sure they apply to your modeling today. Good standards will not be ignored.
Modeling Global User Community President