Fabric is a self-hosted Change Data Capture (CDC) platform that streams changes from your operational databases into a data warehouse in real time. Point it at a database, choose the tables, and every insert, update, and delete lands in your destination within seconds — Apache Iceberg on S3-compatible storage by default, or Delta Lake, S3 Tables, Snowflake, ClickHouse, or a live Postgres mirror.
The key difference: Fabric runs the CDC readers in-process as native connectors. There is no Kafka, no Schema Registry, no Kafka Connect, and no Debezium anywhere in the stack.
How it works
Fabric is the control plane. For each database source you add, it launches one supervised streaming process. That process takes an initial snapshot of the source tables, then streams ongoing changes continuously into your destination.
The heavy lifting stays out of your way. Fabric manages lifecycle, state, and monitoring, so a pipeline is add source → pick tables → start replication.
Everything is pluggable. Sources and destinations are registered behind a common connector SDK, so adding a new database (ingest) or a new warehouse (write) is implementing one interface — no rewiring of the pipeline.
What you get
| Layer | Today |
|---|---|
| Sources | PostgreSQL, MySQL, SQL Server, MariaDB, MongoDB, IBM Db2, or Zoho CRM |
| Destinations | Apache Iceberg (default), Delta Lake, S3 Tables, Snowflake, ClickHouse, Postgres |
| Querying | Trino or any Iceberg engine — one SELECT away |
| Management | Web UI + REST API: sources, syncs, status, logs, reset |
| Streaming stack | None required |
Next steps
Ready to try it? Stand Fabric up in a few minutes with Quickstart, then learn how to create your first pipeline in Setting up a pipeline.
If your source is Postgres or MySQL, one of the three combination guides walks the whole path — destination, source, tables, start, rows arriving — for the sink you have in mind:
- Postgres or MySQL → ClickHouse — the fastest one to stand up.
- Postgres or MySQL → Snowflake — key-pair auth, internal stage,
COPY INTO. - Postgres or MySQL → Iceberg on S3 — the default destination, in your own bucket.