NoSQL Databases

📦 Introduction: The Vending Machine vs. The Custom Basket

Imagine buying snacks in two different ways:

• The Vending Machine (SQL): Everything is strictly catalogued. Slot A1 contains Snickers, A2 contains Twix. You can only get snacks by putting in coins and typing the code. You cannot put a Snickers, a sandwich, and a toy inside the same slot.
• The Custom Storage Basket (NoSQL): You grab a giant plastic bin. You throw in a bag of chips, a book, a jacket, and a customized toy. There are no slots or dividers. You just throw in whatever fits!

In databases, NoSQL represents this custom basket! It is a class of non-relational databases that store data without using rigid table-and-column grids. They are designed to scale horizontally across hundreds of servers easily.

---

🛠️ The Four Kinds of NoSQL

NoSQL databases are grouped into four main architectural patterns:

KEY-VALUE (Redis) DOCUMENT (MongoDB) +-----------------------+ +-----------------------+ | Key: "User-42" | | { | | Val: "Bob, 25, USA" | | "name": "Bob", | +-----------------------+ | "skills": ["SQL"] | | } | +-----------------------+ COLUMN-FAMILY (Cassandra) GRAPH (Neo4j) +-----------------------+ +-----------------------+ | User-42: | | [Alice] --(Friend)--> | | - name: "Bob" | | | | - age: 25 | | v | +-----------------------+ | [Bob] |

1. Key-Value Stores (The Fast Map)

Stores data as simple key-value pairs (like a hash map).

• Primary use: Caching, session storage.
• Example: Redis, Memcached.

2. Document Databases (The JSON Box)

Stores data as structured documents (usually JSON). Each row can have completely different fields!

• Primary use: E-commerce catalogs, user profiles.
• Example: MongoDB, CouchDB.

3. Column-Family (Wide-Column) Stores

Stores columns of data grouped together on disk instead of rows. Excellent for writing massive log files.

• Primary use: Time-series data, heavy write logging.
• Example: Cassandra, ScyllaDB.

4. Graph Databases (The Social Network)

Stores data as Nodes (objects) and Edges (relationships between them). Great for pathfinding.

• Primary use: Social networks, recommendation engines, fraud detection.
• Example: Neo4j.

---

💻 Code Examples: Document Storage Simulation

Let's simulate a NoSQL Document collection storing arbitrary JSON objects in code.

Multi-Language Execution

---

⚠️ Common Mistakes

1. Using NoSQL for highly relational data

Choosing MongoDB when your app requires deep queries joining 5 different tables. Because NoSQL doesn't support JOINs efficiently, you have to write slow nested queries in your application code, causing N+1 query lag.

2. Assuming NoSQL has ACID Transactions

Assuming Cassandra or Redis supports standard transaction rollbacks natively. Many NoSQL databases prioritize speed and scale, sacrificing ACID rules for Eventual Consistency (BASE transactions).

---

🔍 Interview Corner

Q1: When should you choose a NoSQL database over a Relational SQL database?

Choose NoSQL when your data has no strict schema (e.g. log streams, variable product properties), when you need to scale horizontally across hundreds of servers easily, or when you require extremely high write speeds for simple key-value datasets.

Q2: What are the four main types of NoSQL databases?

1. Key-Value Stores (e.g., Redis)
2. Document Databases (e.g., MongoDB)
3. Column-Family / Wide-Column (e.g., Cassandra)
4. Graph Databases (e.g., Neo4j)

---

📝 Summary

• NoSQL databases are non-relational, horizontal-scaling, and schema-free.
• The 4 categories are Key-Value, Document, Column-Family, and Graph.
• They trade strict ACID rules and JOIN operations for blazing write speeds and scale.