Key Characteristics:#

• Schema Flexibility: No more pre-defining every column. Records can have different fields, which is perfect for evolving startups.

• Horizontal Scalability: Instead of buying a “bigger” server (Vertical), you just add “more” cheap servers (Horizontal) to the cluster.

• Distributed Design: Data is replicated across multiple nodes, ensuring that if one server fails, the system stays online.

• BASE over ACID: While SQL prizes strict accuracy (ACID), NoSQL often prizes speed and availability (BASE).

ACID (Used in Relational Databases)#

• ACID ensures that every transaction is handled with full accuracy:

Atomicity: Either everything happens or nothing happens Consistency: Data remains valid before and after operations Isolation: Transactions don’t interfere with each other Durability: Once saved, data is permanent

• This is essential for systems like banking, where mistakes are unacceptable.

BASE (Used in NoSQL Databases)#

• BASE is more relaxed and focuses on performance:

Basically Available: The system always responds Soft State: Data may not be immediately consistent Eventually Consistent: All data will become consistent over time

• This approach is acceptable for systems where immediate accuracy is not critical.

C: Consistency#

Every user sees the same data at the same time.

A: Availability#

Every request gets a response, even if it is not the latest data.

P: Partition Tolerance#

The system continues working even if communication between servers fails.

Types of NoSQL Databases#

Different NoSQL databases are designed for different types of data and use cases.

1. Key-Value Stores

• It Like a giant dictionary. when we provide a Key, and get a Value.
• It is best for Caching, user sessions, shopping carts.
• Examples: Redis, Amazon DynamoDB.

2. Document Databases

• In this database the data are stores as JSON-like documents. We can nest data (like an address inside a user profile).
• It is best for Content Management (CMS), E-commerce catalogs.
• Examples: MongoDB, CouchDB.

3. Column-Family / Wide-Column

• It is Optimized for massive amounts of data across many columns.
• It is best for IoT sensor data, logging billions of events, web indexing.
• Examples: Cassandra, HBase.

4. Graph Databases (The Relationship Experts)

• It Stores “Nodes” (people/things) and “Edges” (relationships).
• It used for Social networks, recommendation engines (e.g., “People you may know”).
• Examples: Neo4j, Amazon Neptune.

5. Time-Series Databases (The Chronologers)

• It is Optimized for data measured over time.
• it is best for Stock market prices, CPU monitoring, weather data.
• Examples: InfluxDB, Prometheus.

6. Vector Databases (The AI Brain)

• It Stores data as mathematical coordinates (embeddings) to find “similar” things.
• It is best for AI/LLM memory, image similarity search, semantic search.
• Examples: Pinecone, Milvus.

SQL vs. NoSQL Comparison Table#

Choosing the Right Database#

If we are building a system, we should ask for these three questions:

• Data Shape: Is it a simple list (Key-Value), a complex object (Document), or a web of connections (Graph)?

• Query Pattern: Will you search by ID, or do you need to do deep math across billions of rows?

• Reliability: Does every millisecond of data need to be 100% accurate (CP), or is being “mostly right” and “always online” better (AP)?

Polyglot Persistence#

Modern applications rarely rely on a single database.Instead, multiple databases are used for different purposes.

Example:

• SQL: Payments (accuracy is critical)
• Redis: Caching and fast access
• MongoDB: User data

This approach is known as polyglot persistence, meaning using multiple database types to solve different problems efficiently.

NoSQL is not a replacement for traditional databases. It is simply a different approach designed for modern challenges.

• SQL: structured, accurate, strict
• NoSQL: flexible, fast, scalable