Embucket: Snowflake compatible lakehouse platform

Embucket is an Apache‑2.0‑licensed, Snowflake‑compatible lakehouse platform built for radical simplicity and full openness. It delivers:

• A Snowflake‑style REST API and SQL dialect
• Apache Iceberg table format under the hood
• Iceberg REST API for both internal and external engines
• Zero‑disk, object‑store‑only architecture (S3 or memory)
• Statically linked single‑binary for effortless deployment
• SlateDB for metadata persistence
• Apache DataFusion as the query engine

Features

• Snowflake compatible API

  • Snowflake SQL syntax dialect
  • Snowflake v1 wire compatible REST API

• Apache Iceberg

  • Data stored in Iceberg format on object storage
  • Built-in internal catalog
  • Exposes Iceberg REST Catalog API

• Zero disk architecture

  • All state (data + metadata) lives in your s3 buckets
  • No other dependencies required

• Scalable "query-per-node" parallelism

  • Spin up multiple Embucket instances against the same bucket
  • Each node handles queries independently for horizontal scale

• Single statically linked binary

  • One embucket executable with zero external dependencies

• Iceberg catalog federation

  • Connect to external Iceberg REST catalogs
  • Read/write across catalogs

Architecture

Embucket is designed with radical simplicity in mind: it is a single binary that runs as a server and provides a REST API for interacting with the lakehouse. It has single dependency - object storage - for both data and metadata.

It is built on top of several open source projects:

• Apache DataFusion - query engine
• Apache Iceberg - data storage
• SlateDB - metadata persistence

Embucket has deep integration with AWS S3 table buckets and relies on them for proper table maintenance.

SLT coverage

These visualizations are automatically updated by CI/CD when tests are run.

DBT Gitlab run results:

This visualization is automatically updated by CI/CD daily

Install Embucket

# Clone and build the Embucket binary
git clone [email protected]:Embucket/embucket.git
cd embucket/
cargo build

Configure and run Embucket

You can configure Embucket via CLI arguments or environment variables:

# Create a .env configuration file
cat << EOF > .env
# SlateDB storage settings
OBJECT_STORE_BACKEND=memory
FILE_STORAGE_PATH=data
SLATEDB_PREFIX=sdb

# Optional: AWS S3 storage (leave blank if using local storage)
AWS_ACCESS_KEY_ID="<your_aws_access_key_id>"
AWS_SECRET_ACCESS_KEY="<your_aws_secret_access_key>"
AWS_REGION="<your_aws_region>"
S3_BUCKET="<your_s3_bucket>"
S3_ALLOW_HTTP=

# Iceberg Catalog settings
# Set to your catalog url
CATALOG_URL=http://127.0.0.1:3000

# Optional: CORS settings
CORS_ENABLED=true
CORS_ALLOW_ORIGIN=http://localhost:8080

EOF

# Load environment variables (optional)
export $(grep -v '^#' .env | xargs)

# Start Embucket
./target/debug/bucketd

Once embucket is running, open:

• localhost:8080 → UI Dashboard
• localhost:3000/catalog → Iceberg REST Catalog API

Demo: running dbt project with Embucket

This demo showcases how to use Embucket with dbt and execute the snowplow_web dbt project, treating Embucket as a Snowflake-compatible database.

Prerequisites

• python 3.9+ installed
• virtualenv installed

Run dbt workflow

# Clone the dbt project with Snowplow package installed
git clone [email protected]:Embucket/compatibility-test-suite.git
cd compatibility-test-suite/dbt-snowplow/

# Set up a virtual environment and install dependencies
virtualenv .venv
source .venv/bin/activate
pip install dbt-core dbt-snowflake

# Activate virtual environment
source .venv/bin/activate

# Set Snowflake-like environment variables
export SNOWFLAKE_USER=user
export SNOWFLAKE_PASSWORD=xxx
export SNOWFLAKE_DB=snowplow
export SNOWFLAKE_SCHEMA=public
export SNOWFLAKE_WAREHOUSE=snowplow

# Install the dbt Snowplow package
dbt deps

# Upload source data
python3 upload.py

# Upload initial data
dbt seed

# Run dbt transformations
dbt run -m snowplow_web

Debugging

Check TRACING_AND_PROFILING.md for information on how to trace and profile embucket service.

Contributing

We welcome contributions! To get involved:

1. Fork the repository on GitHub
2. Create a new branch for your feature or bug fix
3. Submit a pull request with a detailed description

For more details, see CONTRIBUTING.md.

License

This project is licensed under the Apache 2.0 License. See LICENSE for details.