A governed source in open formats.
Business users maintain the data. You read Delta tables in your own bucket, with schema, keys and references enforced on write.
No export jobs. No copies held elsewhere.
Spark, Databricks, Fabric, Trino, DuckDB, Polars and pandas read the tables directly.
import polars as pl path = "s3://lake/commercial/product_master/products" current = pl.read_delta(path) release = pl.read_delta(path, version=41) # any earlier version
SELECT s.order_id, p.category FROM sales s JOIN products_history p ON p.product_id = s.product_id AND s.ordered_at >= p._valid_from AND (s.ordered_at < p._valid_to OR p._valid_to IS NULL)
Built for pipelines, catalogs and BI.
Time travel
Every version kept. Read by version, or ask the API for any date.
SCD type 2 history
One row per version with _valid_from and _valid_to.
Change files
Parquet or CSV copies with the rows changed in each version.
Catalog registration
Unity Catalog or AWS Glue, with descriptions, owner and tags.
dbt seeds
Committed to your Git repository, with tests from the rules.
Flat hierarchies
Level columns, so BI tools group without recursive SQL.
Safe schema changes
Renames and drops need approval. Old names stay for a notice period.
Rules as contracts
Keys, required fields and allowed values, readable as JSON.
API, SDK and CLI
Validate, write to a draft, query with SQL, diff versions.
Point your pipelines at one governed source.
We'll connect it to your lake and catalog in the demo.