Kùzu is an in-process property graph database management system (GDBMS) built for query speed and scalability. Kùzu is optimized for handling complex join-heavy analytical workloads on very large graph databases, with the following core feature set:

• Flexible Property Graph Data Model and Cypher query language
• Embeddable, serverless integration into applications
• Columnar disk-based storage
• Columnar sparse row-based (CSR) adjacency list/join indices
• Vectorized and factorized query processor
• Novel and very fast join algorithms
• Multi-core query parallelism
• Serializable ACID transactions

Kùzu is being actively developed at University of Waterloo as a feature-rich and usable GDBMS. Kùzu is available under a permissible license. So try it out and help us make it better! We welcome your feedback and feature requests.

Precompiled binary of our latest release can be downloaded here.

Our Python package can be directly install through pip.

pip install kuzu

npm install kuzu

We also support official C++ and C bindings. For more installation and usage instructions, please refer to our website.

We take tinysnb as an example graph, which is under dataset/demo-db/csv in our source code, and can be downloaded here.

Start CLI with a database directory ./kuzu_shell ./testdb/

# Data definition
kuzu> CREATE NODE TABLE User(name STRING, age INT64, PRIMARY KEY (name));
kuzu> CREATE REL TABLE Follows(FROM User TO User, since INT64);
# Data loading
kuzu> COPY User FROM "dataset/demo-db/csv/user.csv";
kuzu> COPY Follows FROM "dataset/demo-db/csv/follows.csv";
# Querying
kuzu> MATCH (a:User)-[f:Follows]->(b:User) RETURN a.name, b.name, f.since;

import kuzu

db = kuzu.Database('./testdb')
conn = kuzu.Connection(db)
# Data definition
conn.execute("CREATE NODE TABLE User(name STRING, age INT64, PRIMARY KEY (name))")
conn.execute("CREATE REL TABLE Follows(FROM User TO User, since INT64)")
# Data loading
conn.execute('COPY User FROM "user.csv"')
conn.execute('COPY Follows FROM "follows.csv"')
# Run a query.
results = conn.execute('MATCH (u:User) RETURN COUNT(*);')
while results.has_next():
  print(results.get_next())
# Run a query and get results as a pandas dataframe.
results = conn.execute('MATCH (a:User)-[f:Follows]->(b:User) RETURN a.name, f.since, b.name;').get_as_df()
print(results)
# Run a query and get results as an arrow table.
results = conn.execute('MATCH (u:User) RETURN u.name, u.age;').get_as_arrow(chunk_size=100)
print(results)

// Import library
const kuzu = require("kuzu");

(async () => {
  // Create an empty database and connect to it
  const db = new kuzu.Database("./test");
  const conn = new kuzu.Connection(db);

  // Data definition
  await conn.query(
    "CREATE NODE TABLE User(name STRING, age INT64, PRIMARY KEY (name))"
  );
  await conn.query("CREATE REL TABLE Follows(FROM User TO User, since INT64)");

  // Data loading
  await conn.query('COPY User FROM "user.csv"');
  await conn.query('COPY Follows FROM "follows.csv"');

  // Run a query
  const queryResult = await conn.query("MATCH (u:User) RETURN u.name, u.age;");

  // Get all rows from the query result
  const rows = await queryResult.getAll();

  // Print the rows
  for (const row of rows) {
    console.log(row);
  }
})();

Refer to our Data Import and Cypher section for more information.

To build from source code, Kùzu requires Cmake(>=3.11), Python 3, and a compiler that supports C++20.

• Perform a full clean build without tests and benchmark:
  • make clean && make release
• Perform a full clean build with tests and benchmark (optional):
  • make clean && make all
• Run tests (optional):
  • make test && make pytest

For development, use make debug to build a non-optimized debug version. To build in parallel, pass NUM_THREADS as parameter, e.g., make NUM_THREADS=8.

After build, our CLI binary kuzu_shell is available under the directory build/release/tools/shell/.

We welcome contributions to Kùzu. If you are interested in contributing to Kùzu, please read our Contributing Guide.

By contributing to Kùzu, you agree that your contributions will be licensed under the MIT License.

If you are a researcher and use Kùzu in your work, we encourage you to cite our work. You can use the following BibTeX citation:

@inproceedings{kuzu:cidr,
  author =  {Xiyang Feng and
             Guodong Jin and
             Ziyi Chen and
             Chang Liu and
             Semih Saliho\u{g}lu},
  title={K\`uzu Graph Database Management System},
  booktitle={CIDR},
  year={2023}
}
@misc{kuzu-github,
  author =  {Xiyang Feng and
             Guodong Jin and
             Ziyi Chen and
             Chang Liu and
             Semih Saliho\u{g}lu},
  title = {{K\`uzu Database Management System Source Code}},
  howpublished = {\url{https://github.com/kuzudb/kuzu}},
  month        = nov,
  year         = 2022
}

You can contact us at [email protected] or join our Slack workspace.