Schema registry

A catalog of every dataset and its versions.

The Schema Registry is a catalog of the dataset shapes BARRY has landed. As transports run, the runtime records the schema of each destination dataset — keyed by its fully-qualified name (FQDN) — so you get a versioned, searchable history of every table's columns, types and structure over time.

It is an optional feature: an administrator enables it under Settings → Configuration (the master switch). While off, the Schema Registry sidebar entry and its API are hidden; existing data is untouched.

What it captures

• Datasets — one entry per destination dataset (FQDN), with its owner, description, the transports that contribute to it, when it was last written, and whether it currently has a PII label.
• Versions — each time a dataset's schema changes, a new version is captured. A version lists its columns with data type, nullability, key/partition markers and any PII entities detected (see Presidio).
• Contributors — the transports that write into a dataset, with first/last-seen timestamps (a dataset can be fed by several transports).

What you can do

• Browse & search the dataset catalog.
• Compare versions — a column-level diff between any two versions (added, removed, type-changed, nullability-changed).
• Export a version as JSON, or as DDL, and bulk-export a selection.
• Curate (Operator) — set a dataset's owner, description and column metadata, and activate / deactivate entries. Deleting a version or a dataset is Admin-only.

Why it's useful

• Schema drift — see exactly when and how a source's shape changed, and which downstream datasets are affected.
• Data dictionary — a living description of what each landed dataset contains, with owners and per-column notes.
• Governance — a durable record of structure and PII classification across the estate, independent of the source systems.