RocksDB
rocksdb.org
- Free plan
- Windows
- macOS
- Linux
- iPhone and iPad
- API
- Self-hosted
What RocksDB is
RocksDB is an embeddable C++ key-value store for applications that need persistent, fast storage. It stores keys and values as arbitrary byte arrays, ordering keys with a comparator chosen by the user. Its log-structured merge engine has configurable tradeoffs among write, read, and space amplification, and the project describes it as optimized for low-latency storage on flash drives and high-speed disks. The library supports opening and closing databases, reading and writing, merging, and compaction filters. Its stated uses include database storage engines, application data caches, and embedded workloads. Multi-threaded compactions support databases holding multiple terabytes; the project describes application-server use with up to a few terabytes on local or remote storage. RocksDB is free and open source under either GPLv2 or Apache 2.0, at the user's option. A Java wrapper, RocksJava, is available, with artifacts published through Maven. Building requires GCC 11 or later or Clang 10 or later for C++20 support, and stored values must be smaller than 4 GB.
Who it suits
RocksDB suits developers building database storage engines, application caches, or embedded workloads that need persistent key-value storage. It may also fit application servers managing up to a few terabytes of data.
Good to know, and what to watch
- Free and open source under GPLv2 or Apache 2.0.
- Java wrapper and Maven-published artifacts are available.
- Supports multi-threaded compactions for multi-terabyte databases.
- Optional compression libraries include LZ4 and Zstandard.
- Stored values must be smaller than 4 GB.
- Building requires GCC 11+ or Clang 10+ for C++20.
- Data synchronization is not supported.
Plans
Every plan and priceRocksDB
Open source
Getting started
- Open the RocksDB website
- Choose a supported target such as Linux, macOS, Windows, or iOS
- Prepare GCC 11 or later or Clang 10 or later for C++20 support
- Build the C++ library and select any optional compression libraries
- Use the API to open a database and perform reads and writes
Limits to check
A stored value must be smaller than 4 GB. The library does not provide data synchronization.
The verdict
RocksDB provides a configurable persistent key-value engine with uses ranging from embedded applications to multi-terabyte databases. Check its build requirements and value-size limit against your project before adopting it.
Questions people ask
Is RocksDB free?
Yes. The RocksDB plan is free, and the project is open source.
Which licenses are available?
RocksDB can be used under either GPLv2 or Apache 2.0, at the user's option.
Which platforms are supported?
Supported platforms include Linux, macOS, Windows, iOS, FreeBSD, OpenBSD, AIX, and Solaris. The platform requirements page lists Linux, macOS, and Windows build requirements.
Can Java applications use RocksDB?
Yes. RocksJava is the Java wrapper, and Java artifacts are published through Maven.
What compression integrations are listed?
The build instructions list zlib, bzip2, LZ4, Snappy, and Zstandard as optional compression libraries.
What the maker publishes
- Build requirement
- The project says it depends on GCC 11 or later or Clang 10 or later for C++20 support. source
- Data model
- Keys and values are arbitrary byte arrays, and keys are ordered using a user-specified comparator. source
- Data model
- Keys and values can be arbitrarily sized byte streams. source
- Engine
- RocksDB uses a log-structured merge database design with configurable tradeoffs among write, read, and space amplification. source
- Integrations
- The build instructions list optional compression libraries zlib, bzip2, LZ4, Snappy, and Zstandard. source
- Integrations
- The project documents a MyRocks database storage engine as one example of using RocksDB. source
- Java support
- RocksDB has a Java wrapper called RocksJava, and Java artifacts are published through Maven. source
- Licensing
- RocksDB is available under either GPLv2 or Apache 2.0, at the user's option. source
- Licensing
- RocksDB is dual-licensed under GPLv2 and Apache 2.0, and users may choose either license. source
- Notable limit
- The basic operations documentation says value size must be smaller than 4 GB. source
- Operations
- The library supports database open and close, reads and writes, merging, and compaction filters. source
- Operations
- It provides database open and close, read and write, merge, and compaction filter operations. source
- Performance
- The project describes RocksDB as optimized for low-latency storage such as flash drives and high-speed disks. source
- Performance
- It is optimized for low-latency storage such as flash and high-speed disk drives, and can use read/write rates offered by flash or RAM. source
- Platforms
- Build instructions cover Linux, macOS, FreeBSD, OpenBSD, iOS, Windows, AIX, and Solaris Sparc. source
- Platform support
- The platform requirements page lists Linux, macOS, and Windows build requirements. source
- Product
- RocksDB is an embeddable persistent key-value store for fast storage. source
- Scale
- The project says multi-threaded compactions make RocksDB suitable for storing multiple terabytes in a single database. source
- Scale and audience
- The project describes RocksDB as designed for application servers storing up to a few terabytes on local or remote storage systems. source
- Security reporting
- The security policy asks users to report security issues through Meta's Bug Bounty program instead of public GitHub issues or pull requests. source
- Storage design
- It uses a log-structured database engine written entirely in C++. source
- Support
- The project directs users to GitHub issues for bugs and feature requests and to its Facebook group for questions and discussion. source
- Support
- The project directs users to its Google Group or Facebook Group for general questions and reserves GitHub issues for bug reports. source
- Use cases
- The project describes use in database storage engines, application data caching, and embedded workloads. source
Checked against the maker's pages on 4 Oct 2026.