Lesson 6 — SQL vs NoSQL: Choosing the Right Store

Relational DB

Tables, rows, SQL, strong schema and ACID guarantees.

Document store

Stores JSON-like documents with flexible structure.

Key-value store

Fast get/set by unique key; used for caches and sessions.

Graph database

Stores nodes and edges; queries traverse relationships fast.

Eventual consistency

Replicas converge over time rather than instantly.

1. 1Bank ledger: Relational — transactions and ACID are non-negotiable.
2. 2Game leaderboard: Redis sorted sets — millisecond ranking updates.
3. 3Product catalogue with varied attributes: Document store — every category has different fields.
4. 4'People you may know': Graph database — multi-hop relationship traversal.
5. 5IoT sensor readings: Time-series database — huge append-only writes, time-window queries.

sql-- SQL (PostgreSQL)
SELECT name FROM player WHERE score > 9000 ORDER BY score DESC LIMIT 10;

-- MongoDB aggregation
db.player.find({ score: { $gt: 9000 } }).sort({ score: -1 }).limit(10)

-- Cypher (Neo4j): friends-of-friends, awkward in SQL
MATCH (me:Player {name:'Ana'})-[:FRIEND*2]->(fof)
WHERE NOT (me)-[:FRIEND]->(fof)
RETURN DISTINCT fof.name LIMIT 10;

Simple filtering looks similar everywhere. The difference shows up on relationship-heavy questions, where Cypher stays short and SQL needs repeated self-joins.