Skip to content

Floe

Floe is a policy-based table maintenance system for Apache Iceberg.

Floe automates table maintenance through declarative policies:

  • Compact small files for better query performance
  • Expire old snapshots to reclaim storage
  • Remove orphan files not referenced by any snapshot
  • Optimize manifests for faster query planning

Floe orchestrates maintenance operations that wrap around Iceberg's maintenance procedures.

Quick Start

git clone https://github.com/nssalian/floe.git
cd floe
make start

Open http://localhost:9091

Documentation

Section Description
Configuration Configure catalogs, engines, storage
Architecture System design and extension points

Catalogs

Catalog Description
REST Iceberg REST Catalog
Nessie Git-like versioning
Polaris Multi-engine access control
Hive Hive Metastore
Lakekeeper Open-source Iceberg catalog
Gravitino Unified metadata lake

Engines

Engine Description
Spark Apache Spark (via Livy)
Trino Trino SQL

API

API documentation (once service is up and running): Swagger UI