Which database should I choose for my system?

It depends on your use case. SQL is good for structured data and transactions, while NoSQL is better for scalability and flexible schemas.

Can I use multiple databases in one system?

Yes. Many systems use polyglot persistence where different databases serve different needs.

Databases - System Design

Why It Matters

Database is the core of your system.

Every request eventually reads or writes data.

If your database design is wrong:

system becomes slow
scaling becomes difficult
data inconsistencies happen

Most system design decisions revolve around database choices.

What This Concept Actually Means

A database stores and manages application data.

Different databases are optimized for different use cases.

Choosing the right one is critical.

How It Works

Let’s take an e-commerce system.

Users place orders
Products are listed
Payments are processed

Different parts of system need different databases.

Example:

Orders -> SQL database
Product catalog -> NoSQL database
Search -> Search engine

Flow:


Diagram100%
flowchart LR
  User --> API
  API --> SQLDB
  API --> NoSQLDB
  API --> SearchEngine
Zoom or open fullscreen for a closer look
visualized by

Key Techniques / Variations

1. Relational Databases (SQL)

Examples:

MySQL
PostgreSQL

Characteristics:

structured schema
strong consistency
supports transactions

Use cases:

banking systems
order management

Example:

1,000 transactions per second with ACID guarantees

2. NoSQL Databases

Types:

Key Value Stores

Examples:

Redis

Use cases:

caching
session storage

Document Databases

Examples:

MongoDB

Use cases:

flexible schemas
user profiles

Column Family

Examples:

Cassandra

Use cases:

large scale write-heavy systems

Example:

handles millions of writes per second

Graph Databases

Examples:

Neo4j

Use cases:

social networks
recommendation systems

3. Search Databases

Examples:

Elasticsearch

Use cases:

full text search
filtering and ranking

Example:

search results in 50 ms to 100 ms

4. Time Series Databases

Examples:

InfluxDB

Use cases:

metrics
monitoring systems

Trade-offs and Design Decisions

When to use SQL

need transactions
structured data

When to use NoSQL

need scalability
flexible schema

Pros of SQL

strong consistency
reliable transactions

Cons of SQL

harder to scale horizontally

Pros of NoSQL

high scalability
flexible schema

Cons of NoSQL

weaker consistency in some systems

Real-world example

SQL DB handles 2,000 writes per second
Traffic grows to 10,000 writes per second

Solution:

move to sharding or NoSQL

Architecture / Flow Diagram


Diagram100%
flowchart LR
  Client --> API
  API --> SQLDB
  API --> Cache
  API --> NoSQLDB
Zoom or open fullscreen for a closer look
visualized by

Failure Modes To Watch

database bottleneck
slow queries
lack of indexing
data inconsistency in distributed systems
poor schema design

Design Checklist

Ask yourself:

What type of data am I storing?
What is read vs write ratio?
Do I need strong consistency?
What is expected scale? (e.g., 1k vs 1M requests per second)
Do I need multiple databases?

Summary

Database choice impacts scalability and performance
SQL is best for transactions and structured data
NoSQL is best for scale and flexibility
Different databases serve different purposes
Real systems often use multiple databases together