Open table formats like Apache Iceberg have become the backbone of the lakehouse, but streaming data still arrives with fragmented metadata and governance: topics live in messaging systems, schemas live in registries, and tables live in catalogs—often stitched together by connectors and convention.

In this talk, we propose a Streaming Data Catalog pattern for the Iceberg ecosystem: a metadata layer that treats a continuous stream as a first-class catalog object—on par with tables—while maintaining an explicit stream↔table linkage (typically 1:1) so operational producers and analytical engines share a single, governed representation of the same data. We’ll cover the core metadata model (catalog / namespace / stream), the streaming-specific metadata that tables alone don’t capture (retention, offset-to-file mapping, consumer state, schema evolution contracts), and how the catalog becomes the “brain” that keeps protocol gateways stateless and interoperable.
We’ll also discuss practical interoperability: federating with existing schema registries, synchronizing with external metastores/data catalogs, and enabling multi-protocol ingestion (including lightweight HTTP/gRPC ingestion when deploying a full broker stack isn’t desirable). The goal is to spark a community conversation on standardizing streaming semantics around Iceberg tables—without reinventing the lakehouse.