Back to Lance

Integrations

docs/src/integrations/index.md

9.0.04.1 KB
Original Source

Integrations

Because Lance is built on Apache Arrow and exposes a stable Rust core with Python and Java bindings, it can be consumed from query engines, ML frameworks, databases, and catalog systems without copying or converting data.

This page lists the integrations maintained by the Lance community. Some live in this repository as first-party Rust/Python APIs; others are maintained as dedicated subprojects under the lance-format GitHub organization.

Query Engines and SQL

IntegrationDescriptionSource
Apache DataFusionRegister Lance datasets as DataFusion tables and run SQL with predicate/projection pushdown. Available in both Rust and Python.Built-in
Apache SparkRead and write Lance datasets from Spark SQL and DataFrames.lance-format/lance-spark
DuckDBQuery Lance datasets directly from DuckDB.lance-format/lance-duckdb
TrinoFederate Lance into Trino alongside other connectors.lance-format/lance-trino

Machine Learning and AI

IntegrationDescriptionSource
PyTorchUse lance.torch.data.LanceDataset as a torch.utils.data.IterableDataset for training and inference.Built-in
TensorFlowUse lance_tensorflow.from_lance to stream Lance data into tf.data.Dataset pipelines.lance-format/lance-tensorflow
RayDistributed read/write of Lance datasets with Ray Data.lance-format/lance-ray
Hugging FaceConvert and load Hugging Face datasets to and from Lance in a single call.lance-format/lance-huggingface

Databases and Stream Processing

IntegrationDescriptionSource
PostgreSQLRead Lance datasets from PostgreSQL via the pglance extension.lance-format/pglance
Apache FlinkStream data into Lance from Flink jobs.lance-format/lance-flink

Catalogs and Namespaces

IntegrationDescriptionSource
Lance NamespaceSpecification and codegen SDKs (Rust, Python, Java) for catalog-backed Lance tables.lance-format/lance-namespace
Catalog implementationsReference implementations for Apache Hive, Apache Polaris, Apache Gravitino, Unity Catalog, AWS Glue, and others.lance-format/lance-namespace-impls

Other Ecosystem Projects

IntegrationDescriptionSource
Lance GraphCypher-capable graph query engine on top of Lance.lance-format/lance-graph
Lance Data ViewerRead-only web interface for browsing Lance datasets.lance-format/lance-data-viewer
Lance ContextManage multimodal agentic context lifecycle with Lance.lance-format/lance-context

!!! note "Stability"

Subprojects in the [lance-format](https://github.com/lance-format) organization graduate from incubating
status once they meet the project's quality bar (CI, tests, established use cases, community adoption).
Incubating subprojects may have changing APIs — check the project README for the current status. See
[Community Governance](../community#projects) for details on how integrations are organized.

Additional Integrations

If there's an integration you'd like to see, but isn't listed above, please open an issue describing the use case. PRs are always welcome, though it's recommended to alert the maintainers to avoid duplicated work.