Why Is Redis So Fast? Uncover the Secrets Behind Its Performance

Preface

Redis is a key‑value NoSQL database whose values can be strings, hashes, lists, sets, sorted sets, bitmaps, HyperLogLog, and more. It also provides features such as key expiration, publish/subscribe, transactions, Lua scripting, Sentinel, and Cluster. Official benchmarks claim over 100 k QPS.

1. Development Language

Redis is written in C, a language close to the operating system, which contributes to its execution speed. Learning C helps understand OS fundamentals.

2. Pure In‑Memory Access

All data resides in memory; normal operation requires no disk I/O, giving essentially zero‑latency access. Memory latency is about 100 ns. For example, a 3.1 GHz CPU can execute 3.1 × 10⁹ instructions per second, while memory and disk are orders of magnitude slower.

A typical memory hierarchy shows registers → L1/L2 cache (SRAM) → DRAM → disk, each level adding latency.

3. Single‑Threaded Model

Redis runs a single thread, simplifying algorithm implementation and avoiding the overhead of context switches and lock contention. The downside is that a long‑running command blocks other requests.

Other high‑performance servers such as Node.js and Nginx also use a single‑threaded event loop.

4. Non‑Blocking I/O Multiplexing

Traditional blocking I/O suspends a thread until data arrives. Redis uses epoll for I/O multiplexing, allowing a single process to monitor many file descriptors and only act on those that are ready, eliminating unnecessary blocking.

Example: handling 20 TCP sockets with epoll lets the server process only active sockets, avoiding the overhead of per‑connection processes or threads.