If the formal approach to a relational database is an entity relationship diagram, what is the approach to the NOSQL database?

Question

I am new to database design. As I found out, we use the Entity-Relationship Diagram to help create a relational database.

If the formal approach to the relational database is based on ERD, then what is the approach to the NOSQL database?

Answer 1

I think you can very well use the Entity-Relationship diagram. Just because your database does not handle the relationships between your data entities in the same way that the DBMS does, this does not mean that your data has no relationships, but simply that these relationships are implemented in the application, not the (non-R) DBMS. I think that understanding the information that you are going to store in each record in your database is still worth it.

If you need something more adapted, the problem is that "NoSQL" is an umbrella term for many different database architectures . You probably need to determine which NoSQL database you expect to use before deciding on the best design approach. I think that pretty much any of them you will want to understand what attributes your entities have, even if the database you use provides more flexibility than classical SQL DB (for example, variable sets of key / value pairs)

Answer 2

For graphical databases, models of essential relationships (or their object-oriented cousins, class models) are very good: the coincidence is so good that I think of the graph database as how to implement such a model initially.

All other types of databases (including relational), require the developer to "do extra things" to map the model to the database. For instance. for SQL databases, additional material is relatively small (for example, mapping NN relationships to additional tables, workarounds for inheritance). On the contrary, the additional material for hash tables is quite large (serialization of properties, manual relationship management, etc.).

Of course, different graphical databases also differ in how seamless the connection is. I am participating in a graph database called InfoGrid , where the step from the high-level model to the code is automatic using a code generator.

Answer 3

Hashmaps are used in key storages, charts are used for dbs charts ... These models should be quite simple, but it is doubtful if they are as useful as the er model.