That which you do before beginning the Database model diagram? I am talking about the way you make up the Needs, Specifications etc. Use cases is a factor but other things? Some best practice or a guide? As being a self student I wish to observe how it is going at the disposal of professionals?
Make certain you've got a complete listing of needs out of your client. Do your very best to totally understand these needs, it'll help much inside your design should you choose. If you're determining the needs it might be simpler because you will already know of what you ought to do. Getting an intensive grasp of the goal is an essential part.
If there's an apparent a part of your database that'll be the most crucial (a credit card applicatoin within an online application system for example) I'll usually begin with there and exercise one piece at any given time.
Personally I love to draw rough pictures (no matter what is sensible for you, does not need to be the official ERD) of the items I believe the database may be like and revise it to finer amounts of detail.
Don't depend only on written needs. There's no such factor like a complete listing of needs. Speak with the stakeholders, request questions and employ the outcomes of individuals interviews to determine which characteristics have to be modelled, how they are utilised and also to identify the company secrets. Then some data analysis and analysis is generally needed to look for the right data types along with other aspects.
It might be possible to obtain a good first cut of the data model in advance try not to worry if you cannot. Data modelling generally needs to be an iterative, agile process, completed in sensible sized steps like a project evolves (although you will find certainly cases like Data Warehouse design in which the agile approach might be harder to use).
Based on your clientele, it's really a wise decision to possess two data models and 2 diagrams. One model and diagram is perfect for data analysis. Another is perfect for database design.
I've had great results while on an ER (Entity-Relationship) model and diagram for data analysis as well as an RDM (Relational Data Model) model and diagram to mirror database design.
The ER diagram is helpful for interacting the needs discovered to date to the clients, and ensuring they're complete and proper. ER diagrams are simple to understand even when the customer doesn't have background in database theory. As others have responded, it is really an iterative process, not really a once only waterfall.
The RDM model and diagram is helpful for reflecting logical database design choices like the decision to normalize data or make a move else. It's very easy to derive an RDM model from an ER model, although you need to toss in some design choices which are deliberately overlooked in the ER diagram.
Consequently, it's very easy to construct a table create script from an RDM diagram. You'll have to then add physical features like indexes, to be able to obtain good performance without tearing hair out.