NoSQL (Not Only SQL) databases handle unstructured/semi-structured data at massive scale. Unlike rigid relational tables, NoSQL offers flexible schemas for big data, real-time apps. Popular: MongoDB, Cassandra, Redis. Born from web 2.0 needs (Facebook, Google).

What is NoSQL Database?

NoSQL = Non-relational databases prioritizing:

• Scalability (horizontal - add cheap servers).

• Flexibility (schema-on-read).

• Performance (for massive reads/writes).

Key Difference: No fixed tables/keys—data as JSON docs, key-value pairs, graphs.

Real-life analogy: NoSQL is a flexible warehouse with mixed boxes vs relational's rigid filing cabinets.

Why NoSQL? (Problems with RDBMS)

text
RDBMS Struggles:
- Joins kill performance at scale
- Schema changes = downtime
- Vertical scaling = $$ hardware
- ACID too slow for 1M writes/sec

NoSQL Solves:
- Horizontal scaling (add servers)
- Eventual consistency (BASE vs ACID)
- Handle variety (text, images, logs)

Types of NoSQL Databases (4 Main Categories)

1. Document Stores

Data: JSON/BSON documents (flexible schema).
Use: Content management, e-commerce catalogs.

text
MongoDB Example:
{
  "_id": 1,
  "name": "iPhone 15",
  "price": 79900,
  "specs": {
    "ram": "8GB",
    "camera": "48MP"
  },
  "reviews": ["Great!", "Battery ok"]
}

Examples: MongoDB, CouchDB.

2. Key-Value Stores

Data: Simple key → value (fastest lookups).
Use: Caching, sessions, leaderboards.

text
Redis Example:
SET user:101 '{"name":"Aman","score":8500}'
GET user:101 → Returns JSON

Examples: Redis, DynamoDB.

3. Column-Family (Wide-Column) Stores

Data: Columns grouped by row key (for analytics).
Use: Time-series, logs, recommendations.

text
Cassandra Example:
RowKey: user_101
  - purchases: [item1,time1], [item2,time2]
  - views: [page1,time1], [page2,time2]

Examples: Cassandra, HBase.

4. Graph Databases

Data: Nodes + edges (relationships).
Use: Social networks, fraud detection.

text
Neo4j Example:
(Aman)-[:FRIENDS_WITH]->(Riya)
(Riya)-[:WORKS_AT]->(Google)
Query: Aman's friends at Google?

Examples: Neo4j, JanusGraph.

NoSQL vs SQL: Complete Comparison

CAP Theorem (NoSQL Reality Check)

In distributed systems, pick 2 of 3:

text
C = Consistency (all see same data)
A = Availability (every request gets response)
P = Partition Tolerance (network splits)

SQL: CA (single server)
NoSQL: AP/CP (distributed)

BASE Properties (NoSQL's ACID Alternative)

• Basically Available: Always responds.

• Soft state: Data may change.

• Eventual consistency: Updates propagate eventually.

text
Example: Facebook posts - you see them fast, likes sync in seconds.

Popular NoSQL Databases & Use Cases

Creating NoSQL Example (MongoDB)

javascript
// Insert flexible documents
db.products.insertMany([
  {
    name: "Laptop",
    price: 55000,
    specs: { ram: "16GB", storage: "1TB" }
  },
  {
    name: "Phone", 
    price: 25000,
    // No specs? No problem!
  }
])

// Query
db.products.find({ price: { $lt: 30000 } })

When to Choose NoSQL Over SQL?

text
USE NoSQL WHEN:
✅ Unstructured/varied data (logs, IoT)
✅ Massive scale (>1TB, high writes)
✅ Rapid schema changes
✅ Global distribution

STICK TO SQL WHEN:
✅ Complex transactions (banking)
✅ Strict consistency needed
✅ Known schema upfront
✅ <100GB data

Hybrid Approach: Microservices = SQL for orders + NoSQL for recommendations.

Advantages & Limitations

Pros:

• Horizontal scaling (cheap commodity servers).

• Handle schema evolution.

• High throughput.

• Big data friendly.

Cons:

• Eventual consistency risks.

• No standard query language.

• Complex distributed operations.

• Learning curve per type.

Summary

NoSQL = Flexible, scalable databases for modern big/unstructured data. 4 types: Document, KV, Column, Graph. Follows CAP/BASE vs SQL's ACID.