FalkorDB

Ultra-fast, Multi-tenant Graph Database

Powering Generative AI, Agent Memory, Cloud Security, and Fraud Detection

Our goal is to build a high-performance Knowledge Graph tailored for Large Language Models (LLMs), prioritizing exceptionally low latency to ensure fast and efficient information delivery through our Graph Database.

🆕 FalkorDB is the first queryable Property Graph database to leverage sparse matrices for representing the adjacency matrix in graphs and linear algebra for querying.

• Sparse Matrix Representation: Utilizes sparse matrices to represent adjacency matrices, optimizing storage and performance.

• Linear Algebra Querying: Employs linear algebra for query execution, enhancing computational efficiency.

• Property Graph Model Compliance: Supports nodes and relationships with attributes, adhering to the Property Graph Model.

• OpenCypher Support: Compatible with OpenCypher query language, including proprietary extensions for advanced querying capabilities.

Explore FalkorDB in action by visiting the Demos.

To quickly try out FalkorDB, launch an instance using docker:

docker run -p 6379:6379 -p 3000:3000 -it --rm -v ./data:/var/lib/falkordb/data falkordb/falkordb

Then, open your browser and navigate to http://localhost:3000.

You can also interact with FalkorDB using any of the supported Client Libraries

In this example, we'll use the FalkorDB Python client to create a small graph representing a subset of motorcycle riders and teams participating in the MotoGP league. After creating the graph, we'll query the data to explore its structure and relationships.

from falkordb import FalkorDB

# Connect to FalkorDB
db = FalkorDB(host='localhost', port=6379)

# Create the 'MotoGP' graph
g = db.select_graph('MotoGP')
g.query("""CREATE (:Rider {name:'Valentino Rossi'})-[:rides]->(:Team {name:'Yamaha'}),
                  (:Rider {name:'Dani Pedrosa'})-[:rides]->(:Team {name:'Honda'}),
                  (:Rider {name:'Andrea Dovizioso'})-[:rides]->(:Team {name:'Ducati'})""")

# Query which riders represents Yamaha?
res = g.query("""MATCH (r:Rider)-[:rides]->(t:Team)
                 WHERE t.name = 'Yamaha'
                 RETURN r.name""")

for row in res.result_set:
	print(row[0])

# Prints: "Valentino Rossi"

# Query how many riders represent team Ducati ?
res = g.query("""MATCH (r:Rider)-[:rides]->(t:Team {name:'Ducati'})
                 RETURN count(r)""")

print(res.result_set[0][0])
# Prints: 1

You can call FalkorDB's commands from any Redis client. Here are several methods:

$ redis-cli
127.0.0.1:6379> GRAPH.QUERY social "CREATE (:person {name: 'roi', age: 33, gender: 'male', status: 'married'})"

You can interact with FalkorDB using your client's ability to send raw Redis commands.

Note: Depending on your client of choice, the exact method for doing that may vary.

This code snippet shows how to use FalkorDB with from Python using falkordb-py:

from falkordb import FalkorDB

# Connect to FalkorDB
db = FalkorDB(host='localhost', port=6379)

# Select the social graph
g = db.select_graph('social')

reply = g.query("CREATE (:person {name:'roi', age:33, gender:'male', status:'married'})")

Note: Some languages have client libraries that provide support for FalkorDB's commands:

Project	Language	License	Author	Stars	Package	Comment
jfalkordb	Java	BSD	FalkorDB		Maven
falkordb-py	Python	MIT	FalkorDB		pypi
falkordb-ts	Node.JS	MIT	FalkorDB		npm
falkordb-rs	Rust	MIT	FalkorDB		Crate
falkordb-go	Go	BSD	FalkorDB		GitHub
NFalkorDB	C#	Apache-2.0	FalkorDB		nuget

Project	Language	License	Author	Stars	Package	Comment
nredisstack	.NET	MIT	Redis		nuget
redisgraph-rb	Ruby	BSD	Redis		GitHub
redgraph	Ruby	MIT	pzac		GitHub
redisgraph-go	Go	BSD	Redis		GitHub
rueidis	Go	Apache 2.0	Rueian		GitHub
ioredisgraph	JavaScript	ISC	Jonah		GitHub
@hydre/rgraph	JavaScript	MIT	Sceat		GitHub
php-redis-graph	PHP	MIT	KJDev		GitHub
redisgraph_php	PHP	MIT	jpbourbon		GitHub
redisgraph-ex	Elixir	MIT	crflynn		GitHub
redisgraph-rs	Rust	MIT	malte-v		GitHub
redis_graph	Rust	BSD	tompro		GitHub
rustis	Rust	MIT	Dahomey Technologies		Crate	Documentation
NRedisGraph	C#	BSD	tombatron		GitHub
RedisGraph.jl	Julia	MIT	xyxel		GitHub

Official Docs | Clients | Commands | 📊 Latest Performance Benchmarks

• Discussions: Join our community discussions on GitHub Discussions to ask questions, share ideas, and connect with other users.

• Contributing: We welcome contributions! See the Developer Guide below to build FalkorDB from source and run the test suites.

• License: This project is licensed under the Server Side Public License v1 (SSPLv1). See the LICENSE file for details.

This repository is the Rust implementation of FalkorDB. It builds the falkordb Redis module (libfalkordb.{so,dylib}) from Rust, using GraphBLAS sparse matrices for graph storage and traversal.

The easiest way to get started is using the development container, which includes all dependencies pre-installed:

1. Install Docker and VS Code
2. Install the Dev Containers extension
3. Open this project in VS Code
4. Click "Reopen in Container" when prompted (or press F1 and select "Dev Containers: Reopen in Container")
5. Wait for the container to build (first time takes ~10-15 minutes)
6. Start developing! All dependencies are ready to use.

See .devcontainer/README.md for more details.

If you prefer to set up the environment manually:

cargo build

GraphBLAS, LAGraph, and RediSearch must be built and installed before building this project.

Local builds use whatever OpenMP package is on the system — build/libomp.sh is not required for local development. It is only invoked by the Docker toolchain image (build/Dockerfile) to produce /opt/libomp/lib/libomp.a, which lets the published libfalkordb.{so,dylib} embed libomp statically and have no libomp.so.5 / libgomp.so.1 / libomp.dylib runtime dependency. A locally-built artifact will dynamically link the system libomp instead — fine for dev, but CI/Docker is the source of truth for the self-contained image.

If you do want a self-contained local artifact (e.g. to mirror the Docker build), run build/libomp.sh with a writable PREFIX and point graph/build.rs at it via LIBOMP_PREFIX:

CC=$(brew --prefix llvm)/bin/clang PREFIX=$HOME/libomp ./build/libomp.sh
LIBOMP_PREFIX=$HOME/libomp cargo build

The script auto-detects the libomp source release from ${CC:-clang} --version, so it stays ABI-matched to your compiler with no manual version arg. In Docker the same auto-detection runs against clang-${CLANG_MAJOR}, eliminating the prior drift risk between the apt-installed clang and a hand-pinned LLVMORG_VERSION.

GraphBLAS, LAGraph and RediSearch are git submodules under deps/, built by the native-deps crate. git owns those checkouts — native-deps only builds what is there, so populate them first.

On macOS, point the build at Homebrew clang first: the system clang has no OpenMP, and GraphBLAS would build single-threaded without saying so.

export CC=$(brew --prefix llvm)/bin/clang
export CXX=$(brew --prefix llvm)/bin/clang++

git submodule update --init --recursive
cargo build   # graph/build.rs calls native-deps; nothing else to run

(git clone --recurse-submodules does the first step for you.)

cargo build builds any missing dependency automatically, so the explicit step below is only needed if you want to build them ahead of time or inspect the result:

cargo run --manifest-path native-deps/Cargo.toml

Results are cached at ${XDG_CACHE_HOME:-$HOME/.cache}/falkordb/native-deps/<dep>/<key>/ ($FALKORDB_DEPS_CACHE overrides the root). The key covers everything the artifacts are ABI-tied to: the submodule revision, the GB_control patch, the vendored PreJIT kernels, the recipe sources, $CC/$CXX --version, the target triple and the OpenMP/sanitizer flavour. Worktrees therefore share archives, and a compiler bump can never yield a stale-ABI cache hit.

$FALKORDB_NATIVE_DEPS_PREBUILT names extra read-only roots to search first; the Docker images use it to expose the artifacts their dep stages built.

Move the gitlink the ordinary way, then regenerate the lock file:

git -C deps/RediSearch checkout <sha>
git add deps/RediSearch
cargo run --manifest-path native-deps/Cargo.toml -- lock

deps/native-deps.lock mirrors the gitlinks as a plain file, because a Docker build context has no .git for the dep stages to read. CI enforces that the two agree via lock --check.

• pytest - create virtualenv and install tests/requirements.txt

The virtual environment should be activated before running tests.

python3 -m venv venv
source venv/bin/activate
pip install -r tests/requirements.txt

• run unit tests with cargo test -p graph

• run e2e and function tests with pytest tests/test_e2e.py tests/test_functions.py -vv

• run MVCC and concurrency tests with pytest tests/test_mvcc.py tests/test_concurrency.py -vv

• run flow tests with ./flow.sh

• run tck tests with pytest tests/tck/test_tck.py -s

There is an option to run only part of the TCK tests and stop on the first fail

TCK_INCLUDE=tests/tck/features/expressions/list pytest tests/tck/test_tck.py -s

To run all passing TCK tests use:

TCK_DONE=tck_done.txt pytest tests/tck/test_tck.py -s

Licensed under the Server Side Public License v1 (SSPLv1). See LICENSE.

⭐️ If you find this repository helpful, please consider giving it a star!

↗️ Graph, graph database, RAG, graphrag, Retrieval-Augmented Generation,Information Retrieval, Natural Language Processing, LLM, Embeddings, Semantic Search