# SQL vs NoSQL — System Design

Source: https://www.geekswithgeeks.com/en/system-design/sd-databases

> Choosing a data store by access pattern, not hype.

## Relational strengths

SQL databases give strong schemas, ACID transactions, and flexible joins. Ideal when data is highly related and correctness matters (payments, inventory).

## NoSQL families

**Key-value** (Redis, DynamoDB), **document** (MongoDB), **wide-column** (Cassandra), **graph** (Neo4j). They trade joins and transactions for horizontal scale and flexible schema.

## Filing cabinet vs sticky notes

SQL is a labelled filing cabinet — rigid but you can cross-reference anything. NoSQL is a wall of sticky notes — fast to add, but you must arrange them for how you'll read them.

## Model the query first

With NoSQL you design tables around the exact reads you need. If access patterns are unknown or changing, relational is safer.