Beyond Tables
NoSQL (Not Only SQL) databases are designed for specific use cases where relational databases aren't ideal — massive scale, flexible schemas, or specialized data structures. They don't use SQL as their primary query language.
NoSQL Database Types
Document Store Key-Value Store
┌────────────────────┐ ┌─────────┬───────────┐
│ { │ │ "user:1"│{name:Alice}│
│ "_id": "123", │ ├─────────┼───────────┤
│ "name": "Alice", │ │ "user:2"│{name:Bob} │
│ "age": 25, │ ├─────────┼───────────┤
│ "courses": │ │ "sess:5"│"abc123" │
│ ["DB", "Net"] │ └─────────┴───────────┘
│ } │ Redis, DynamoDB
MongoDB, CouchDB
Column-Family Graph Database
┌─────┬───────┬──────┐ (Alice)──knows──→(Bob)
│ Row │ Col1 │ Col2 │ │ │
├─────┼───────┼──────┤ │ │
│ 1 │ Alice │ 25 │ knows knows
│ 2 │ Bob │ 30 │ │ │
│ 3 │ Carol │ 28 │ (Carol)←knows──(Dave)
└─────┴───────┴──────┘
Cassandra, HBase Neo4j, ArangoDB
When to Use NoSQL
Use NoSQL when: Use SQL when:
───────────────────────────── ─────────────────────────────
✓ Schema is evolving rapidly ✓ Data is highly structured
✓ Massive horizontal scale ✓ Complex queries & JOINs
✓ Real-time analytics ✓ ACID transactions needed
✓ Simple key-value lookups ✓ Data consistency critical
✓ Unstructured data ✓ Reporting & analytics
CAP Theorem
You can only guarantee TWO of three:
Consistency
/\
/ \
/ \
/ CA \
/________\
CP AP
CA: Consistency + Availability (traditional RDBMS)
CP: Consistency + Partition tolerance (MongoDB, HBase)
AP: Availability + Partition tolerance (Cassandra, DynamoDB)
In distributed systems, partition tolerance is required,
so you choose between consistency and availability.