Why Single‑Threaded Redis Is So Fast: Four Key Design Choices

Short Talk

Redis's performance can be attributed to four main factors.

In‑memory storage

Optimized data structures

Single‑threaded architecture

Non‑blocking I/O

In‑Memory Storage

Redis stores key‑value data directly in memory, making every read and write operation a memory access, which is orders of magnitude faster than disk access.

Optimized Data Structures

Redis leverages a variety of specialized data structures to store data efficiently without worrying about persistence.

For example, Redis lists are implemented as linked lists, providing O(1) insertion and deletion at both ends, while sorted sets use skip‑lists for fast lookup and insertion.

Single‑Threaded Architecture

Redis handles reads and writes extremely quickly, and CPU usage is rarely a concern.

According to the official documentation, a single Redis instance on a typical Linux system can process up to one million requests per second.

The main bottleneck is network I/O; most processing time is spent waiting for I/O.

Although multithreading allows concurrent task handling, it adds context‑switch and lock overhead that provides little performance gain for Redis.

Benefits of the single‑threaded design include:

Minimizing CPU cost of thread creation/destruction

Reducing CPU cost of context switches

Eliminating lock overhead and related bugs

Allowing the use of "thread‑unsafe" commands such as LPUSH

Non‑Blocking I/O

To handle incoming requests, the server must perform system calls on sockets, which are typically blocking operations.

I/O multiplexing monitors many sockets simultaneously and returns only those ready for reading.

Ready sockets are pushed to the single‑threaded event loop and processed using a reactive model.

In summary:

Network I/O is slow because it blocks.

Redis can execute commands quickly because they run entirely in memory.

Therefore Redis adopts:

I/O multiplexing to alleviate slow network I/O.

A single‑threaded architecture to reduce lock overhead.

Conclusion

The four reasons—memory storage, optimized structures, single‑threaded design, and non‑blocking I/O—explain why Redis remains one of the fastest and most widely used in‑memory data stores despite being single‑threaded.

This article is translated from a Medium post (original URL: https://levelup.gitconnected.com/4-reasons-why-single-threaded-redis-is-so-fast-414e0106f921).

Backend Community Invitation

Backend Exclusive Technical Group

Build a high‑quality technical community; developers, technical recruiters, and anyone willing to share job referrals are welcome to join and help each other grow.

Civilized discussion focuses on technical exchange , job referrals , and industry exploration .

Advertisements are prohibited; beware of private messages that may be scams.

Contact me to be added to the group.