Introduction to Redis

Redis stands for Remote Dictionary Server. It is an open-source, in-memory data store used as a database, cache, message broker, and streaming engine.

Think of Redis as an ultra-fast notepad that lives in your computer’s RAM. When you need to remember something quickly, instead of going to a library (your database on disk), you look at your notepad (Redis in RAM). It is almost instantaneous.

Analogy: Imagine you are a cashier at a supermarket. You have a drawer next to you with the prices of the 50 most common items — you don’t need to look them up in a catalog every time. That drawer is Redis. The catalog is your database.

Salvatore Sanfilippo was building a real-time analytics tool. His PostgreSQL database could not handle thousands of reads and writes per second fast enough. He needed something that could respond in microseconds, not milliseconds.

He realized:

• Disk I/O (reading from a hard drive) is the bottleneck
• RAM is orders of magnitude faster than disk
• Most frequently accessed data is a small subset of the total data

So he built Redis — a data store that lives entirely in RAM.

Traditional relational databases (MySQL, PostgreSQL) store data on disk. Every query involves:

1. Parsing the SQL query
2. Planning the query execution
3. Reading data from disk (HDD/SSD)
4. Joining tables if needed
5. Returning the result

This process can take 5–50 milliseconds or more for complex queries. Under high load (thousands of requests per second), this becomes a serious bottleneck.

Caching stores the result of an expensive operation so you can reuse it without recomputing it. It is the single most impactful optimization in web application performance.

Example: You have a news website. The homepage shows the top 10 trending articles. This query joins 3 tables, sorts by views, and takes 200ms. With 10,000 users per minute requesting the homepage:

• Without cache: 10,000 × 200ms = database is overwhelmed
• With Redis cache: Query runs once, result cached for 60 seconds, 10,000 requests served from RAM in < 1ms each

1. In-Memory Storage — All data lives in RAM, no disk I/O
2. Single-threaded Event Loop — No context switching overhead between threads
3. Efficient Data Structures — Internally optimized (skip lists, hash tables, zip lists)
4. Non-blocking I/O — Uses multiplexing to handle thousands of connections
5. Simple Protocol (RESP) — Lightweight binary-safe protocol

Redis can handle 1,000,000+ operations per second on a single server.

flowchart TB
    Client[Client / Browser] --> App[Application Server]
    App --> Cache{Check Redis Cache}
    Cache -->|Cache HIT ✅| Fast["Return cached data<br/>< 1ms"]
    Cache -->|Cache MISS ❌| DB[Query Database<br/>PostgreSQL / MySQL<br/>~5-50ms]
    DB --> Store[Store result in Redis<br/>with TTL]
    Store --> Return[Return data to client]
    Fast --> Client
    Return --> Client

    style Client fill:#3b82f6,color:#fff
    style App fill:#7c3aed,color:#fff
    style Cache fill:#f59e0b,color:#fff
    style Fast fill:#059669,color:#fff
    style DB fill:#3b82f6,color:#fff
    style Store fill:#7c3aed,color:#fff
    style Return fill:#10b981,color:#fff

Flow: Client requests data → App checks Redis cache → HIT: serve instantly from RAM (0.1ms) → MISS: query slow database, store result in Redis for next time, then return.