# The NoSQL Families in Plain Words

NoSQL means "not only SQL", the databases that don't keep their data in relational tables; its four families store key-value pairs, documents, wide rows of columns or a graph of connections, and MongoDB is a document database.

Novice · NoSQL · MongoDB · Updated November 19, 2026

By Gregor Dzierzon

## After this lesson you can

- Explain what NoSQL means, and which family MongoDB belongs to
- Match a job to the NoSQL family that fits it, key-value, document, wide-column or graph

## Key takeaways

- NoSQL means "not only SQL". It names the databases that don't keep their data in relational tables.
- A key-value store finds a value by its key, the way a coat check finds a coat by its ticket, and never looks inside the value.
- A document database stores each record as a document it can search inside, like a folder of filled-in forms. MongoDB is one.
- A wide-column database spreads a huge number of rows over many machines, each row with its own columns, such as every page view of every recipe.
- A graph database stores things and the connections between them, like a map of who is friends with whom.

Read a developer job ad, or the menu of any cloud provider, and you'll meet the word NoSQL, usually beside a list of names. Those names belong to a few families, and the family tells you what each one is good at. When an app you build needs a new kind of storage, or a teammate suggests a database you've never used, knowing the families lets you follow the conversation and ask the right question.

Sunday Kitchen is growing, and its website has four new jobs:

1. Remember what's in each visitor's shopping list while they browse, and forget it when they leave.
2. Store its recipe cards, each with lists of ingredients and steps.
3. Record every page view, millions of them, to see which recipes are popular.
4. Suggest "cooks who saved this recipe also saved…", from who saved what.

One kind of database isn't the best tool for all four. That's why NoSQL comes in families.

## What NoSQL means

*Tables and Documents: Two Ways to Store a Recipe* showed the relational way: tables of rows, joined together, with SQL as the language that asks them questions. Relational databases ran nearly everything for decades. Then big websites needed to store far more data, spread across many machines, in shapes that tables handle awkwardly, and new databases appeared that don't use relational tables at all. They got the name **NoSQL**, read today as "not only SQL": another set of tools beside relational databases, not a replacement for them.

NoSQL isn't one design. It's four main families, each built around one everyday idea.

## Key-value and document

**Key-value: a coat check.** You hand over your coat and get ticket 47. Later you hand back ticket 47 and get your coat. The attendant never looks in the pockets. A **key-value store** works the same way: you save a value under a key, such as `cart-381`, and fetch it back by that key, very fast. It can't answer "which carts hold flour?", because it never looks inside a value. That's job 1, the shopping list, and also sign-in sessions and saved copies of busy pages. Redis is the best-known key-value store.

**Document: a folder of filled-in forms.** Each record is a document, a whole form with its own fields and lists, like the recipe card in the last lesson. Unlike a coat check, the database reads inside the forms, so it can answer "which recipes are vegetarian?" or "which use butter?". A **document database** fits job 2, the recipe cards. MongoDB is a document database, and the rest of this course is about it.

## Wide-column and graph

**Wide-column: one enormous ledger, split over many machines.** Picture a logbook so big it's kept in hundreds of volumes, each on its own shelf, with one page per recipe per day. Each row has a key, such as a recipe and a day, and can hold its own set of columns, such as the time of every view. A **wide-column database** (also called a column-family database) is built to take in huge numbers of writes across many machines, as long as you plan its tables around the exact questions you'll ask. That's job 3, the page views, and also sensor readings and chat messages at large scale. Cassandra is the best-known one.

**Graph: a map of who knows whom.** Draw every cook and every recipe as a dot, and a line from a cook to each recipe they saved. "Cooks who saved this also saved…" means following lines: from a recipe to the cooks who saved it, then on to the other recipes they saved. A **graph database** stores the dots and the lines themselves, so following them is fast. That's job 4. In tables, the same question means joining a table to itself again and again. Neo4j is the best-known graph database.

![Four small panels, one for each NoSQL family. Key-value: a coat check, with ticket 47 handed over for one coat. Document: a folder of filled-in recipe forms, one form open. Wide-column: a huge ledger split into many volumes on shelves, one page per recipe per day. Graph: dots for cooks and recipes joined by lines, showing who saved which recipe.](https://fsdtrainingmedia.blob.core.windows.net/public/images/2026/11/96192ecf27274b04b3660fc2f9f253c2.webp)

## Where MongoDB sits

| Family | The everyday picture | Good at | Sunday Kitchen's job | Best-known |
|---|---|---|---|---|
| Key-value | A coat check | Fetching one value by its key, very fast | Each visitor's shopping list | Redis |
| Document | A folder of filled-in forms | Whole records with lists inside, searched by any field | The recipe cards | MongoDB |
| Wide-column | A ledger split over many machines | A huge flood of writes, read by planned keys | Every page view | Cassandra |
| Graph | A map of who knows whom | Following connections | "Cooks who saved this also saved…" | Neo4j |

MongoDB sits in the document family. The lines between families blur at the edges: many products can do a little of another family's job, some cover several, and relational databases can store documents in a column too. Still, each product has a home family, the job it's built for. For a recipe card, an order, a profile or a product page, a whole record with lists inside, that home is documents.

## Check yourself

2 questions to try on the lesson page.

## So what

When someone names a NoSQL database, you can now place it in its family and say what it's good at, and you know why this course reaches for a document database.

## Key terms

- [NoSQL](https://www.ask-codey.com/glossary/nosql)
- [Key-Value Store](https://www.ask-codey.com/glossary/key-value-store)
- [Document Database](https://www.ask-codey.com/glossary/document-database)
- [Column-Family Database](https://www.ask-codey.com/glossary/column-family-database)
- [Graph Database](https://www.ask-codey.com/glossary/graph-database)

## Related

- [Tables and Documents: Two Ways to Store a Recipe](https://www.ask-codey.com/lessons/mongodb-tables-and-documents)
- [Where Documents Fit, and Where They Don't](https://www.ask-codey.com/lessons/mongodb-where-documents-fit)

Source: https://www.ask-codey.com/lessons/mongodb-nosql-families
