Part 2 of 5 · 1 min read
What "NoSQL" actually groups together
A shared label for several unrelated data models, not one single alternative.
Think of the word "vehicle." A bicycle, a cargo ship and a jet are all vehicles, but they share almost nothing about how they are built or used. Calling something "not a car" says far less than naming which kind of vehicle it actually is.
In plain words: "NoSQL" is a loose label for databases that do not use the relational model's fixed-schema tables. It covers several genuinely different data models, grouped together mainly by what they are not, rather than by any one shared design. Four families are common enough to name here: document, key-value, graph, and column-family stores.
A worked example. A document database, a key-value store and a graph database can all be called NoSQL, but a document database's closest relative in this list is not the other two. It is, in many practical ways, closer to a relational table with flexible columns than to a graph database built entirely around relationships.
The key lines. No SQL here deliberately, since this part is about naming the families, not querying any of them yet.
NoSQL
NoSQL describes a database that does not organize data as fixed-schema relational tables. It is not one data model. It is a label covering several different data models that share this one trait.
Example: a key-value store and a graph database are both called NoSQL, despite storing and querying data in almost entirely different ways.