Snapchat

KeyDB: Run a Multithreaded Redis-Compatible Database

KeyDB is a high-performance, multithreaded fork of Redis for workloads that need more throughput or features such as active replication. Its Redis protocol compatibility makes it a potential drop-in replacement, with Linux-focused build and deployment requirements.

View on GitHub Compare
Stars
12.5k
Forks
677
Stars, last 30 days
-1

Our take

Health score

40/100

How it is scored
Activity
19
Community
75
Issues
46
Pull requests
20

-1 stars in the last 30 days

  • More contributors than 78% of the projects we track

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

12,500

Forks

679

Change since Nov 17, 2026: +2 stars·0 forks

Overview

KeyDB is a Redis-compatible in-memory database that runs request handling across multiple threads. It aims to increase throughput and memory efficiency while preserving compatibility with the Redis protocol, modules, and scripts, so existing Redis clients and deployments can be candidates for migration.

It also adds capabilities such as active replication, FLASH storage, and non-blocking KEYS and SCAN queries through its MVCC architecture. It is most relevant to teams operating Redis-compatible workloads that want to evaluate these features or improve throughput on suitable hardware.

Key Features

  • Multithreaded request handling for concurrent network I/O and query parsing.
  • Redis protocol, module, and script compatibility, including transaction and script atomicity guarantees.
  • Active replication, including configurations where multiple instances accept reads and writes.
  • MVCC architecture intended to let KEYS and SCAN run without blocking the database.
  • Optional FLASH storage for SSD-backed data, described as experimental.
  • TLS-enabled builds by default, with an option to build without TLS.

Use Cases

  • Redis operators evaluating a compatible alternative for workloads where request throughput is a concern.
  • Teams that want active-active replication to distribute writes across synchronized instances.
  • Services that rely on Redis clients, scripts, or modules and want to test a compatible database fork.
  • Operators considering SSD-based FLASH storage or S3-backed database dumps, after reviewing the documented caveats.

Getting Started

On Linux, install the documented build dependencies, initialize the submodules, and compile:

git submodule init && git submodule update
make

See the README and build documentation for dependencies, configuration, tests, and installation details.

Alternatives

  • dragonfly: Dragonfly offers Redis and Memcached APIs, while KeyDB is a Redis fork with features such as active replication.
ProjectLanguageLicenseStarsStatus
KeyDBC++BSD-3-Clause12.5kNot checked yet
dragonflyC++–31.8kActive

Considerations

  • The README describes Linux as the tested platform and notes reliance on SO_REUSEPORT, which limits current support to Linux.
  • Multithreaded performance depends on configuration and workload. The project advises tuning server threads in relation to network queues and notes that excessive thread counts can reduce performance.
  • FLASH storage is explicitly experimental and should be used with discretion.
  • The README warns that multi-master-no-forward can cause data loss unless masters are connected in a true mesh topology.
  • Benchmarking requires care: the README says the included benchmark tools are single-threaded and recommends a cluster benchmarking tool such as memtier.

Found this useful?

Share it with someone who would like KeyDB.

Comparisons