Open-source alternatives to Snowflake, Fivetran, and dbt Cloud?

Summary

• Each of the three products covers one layer, and each layer has open-source options: an open table format such as Delta Lake or Apache Iceberg queried by Apache Spark, Trino or DuckDB for the warehouse; Debezium and Kafka Connect for ingestion; and dbt OSS or dbt Core scheduled by Apache Airflow for transformation.
• dbt's names have changed. Since September 2026 dbt Labs has called its full v2 distribution "dbt" and its Apache 2.0 subset "dbt OSS," dbt Core v1.x stays Apache 2.0, and dbt Cloud is now called the dbt platform, a hosted commercial product.
• The managed products' docs spell out what you would take on. Snowflake says there is no hardware and virtually no software for you to install, configure or manage, and Fivetran says it designs, builds, operates and maintains extract and load.
• Check licenses component by component. Airbyte's own connectors are open source but its platform uses the Elastic License 2.0, and dbt's full v2 distribution includes proprietary code, while dbt OSS is Apache 2.0.
• Choose on criteria: license terms, who runs and upgrades each piece, connector coverage for your sources, governance across layers, and whether data stays in an open format that any engine can read.

Open-source alternatives to Snowflake, Fivetran, and dbt Cloud?

There's an open-source option at every layer these three products cover, though no single project covers all three. For storage and SQL, keep tables in an open format such as Delta Lake or Apache Iceberg on object storage and query them with Apache Spark, Trino or DuckDB. For ingestion, capture database changes with Debezium and move them with Kafka Connect. For transformation, run dbt OSS or dbt Core and schedule it with Apache Airflow. The trade is operational: you install, configure, upgrade and operate each piece, which is work the managed products' docs say they handle for you.

What each managed product handles, per its docs

Snowflake. Its architecture has three layers: database storage, compute and cloud services. Snowflake's docs say "there is no hardware (virtual or physical) for you to select, install, configure, or manage."
Fivetran. It follows ELT: extract and load raw data, then transform it in the destination. Each connector creates and manages its own schema, and Fivetran "designs, builds, operates, and maintains the extract and load from the source system to the destination schema," including keeping connectors current as sources change.
dbt Cloud. dbt Labs' docs now call it the dbt platform (formerly dbt Cloud). It runs dbt in a hosted environment with a browser-based interface and adds job scheduling, CI/CD, documentation hosting, monitoring and alerting, and an IDE.

Open-source options by layer

Layer	Managed product	Open-source projects	License (per each project)
Table storage	Snowflake storage layer	Delta Lake, Apache Iceberg	Apache 2.0
SQL engine	Snowflake compute layer	Apache Spark, Trino, Presto, DuckDB	Apache 2.0 (DuckDB: MIT)
Catalog and governance	Snowflake cloud services layer	Unity Catalog OSS	Apache 2.0
Ingestion	Fivetran connectors	Debezium for change data capture; Kafka Connect for streaming data between Kafka and other systems	Apache 2.0
Ingestion (mixed license)	Fivetran connectors	Airbyte	Platform: Elastic License 2.0; Airbyte's own connectors: open source
Transformation	dbt platform	dbt OSS (v2) or dbt Core (v1.x)	Apache 2.0
Scheduling	dbt platform jobs	Apache Airflow	Apache 2.0

Two notes on the table. Debezium is "a set of distributed services to capture changes in your databases," and its source connectors run on Kafka Connect or as a standalone Debezium Server. And on dbt: since September 2026 the full distribution is called dbt and adds features from proprietary code under the dbt Product Licensing Agreement, while dbt OSS is "entirely built on code released under the Apache 2.0 license."

How to choose

1. License terms. Apache 2.0 and MIT allow use, modification and redistribution. The Elastic License 2.0 bars offering the software to third parties as a hosted or managed service. Check every component, not just the headline project.
2. Operations. Someone has to size clusters, apply upgrades, rotate credentials and watch failed runs. Decide who owns that work before you move.
3. Connector coverage. List every source system and confirm a maintained connector exists, including how it handles schema changes.
4. Governance across layers. In a stack built from separate tools, each tool can hold its own permissions. A shared catalog keeps access rules and lineage in one place.
5. Data portability. Open table formats keep your data readable by any engine that implements the format, so changing engines later doesn't mean rewriting the data.

How Databricks approaches this

Databricks is a commercial platform, not an open-source project, but it runs on open formats and engines. Delta Lake, which Databricks originally developed, is its default table format; it runs Apache Spark workloads and supports Apache Iceberg tables; and it governs data with Unity Catalog, which Databricks open-sourced under Apache 2.0. It's designed to cover the same three layers as managed services while the data stays in open formats. SQL warehouses serve the SQL layer. Lakeflow Connect managed connectors produce ingestion pipelines that are governed by Unity Catalog and run on serverless compute. Lakeflow Pipelines extend and interoperate with Apache Spark Declarative Pipelines. Teams that already use dbt can keep it: the dbt task in Lakeflow Jobs runs dbt projects on Databricks with the dbt-databricks adapter.

FAQs

Is dbt Core still open source?

Yes. dbt Labs' licensing FAQ says dbt Core v1.x is still Apache 2.0, and dbt OSS, the open-source subset of dbt v2, is Apache 2.0 as well.

Is Airbyte open source?

Partly. Airbyte's docs say its own connectors remain open source, while the platform is licensed under the Elastic License 2.0, which restricts offering it as a hosted service.

Can I keep dbt if I move to Databricks?

Yes. The dbt task in Lakeflow Jobs runs dbt projects with the dbt-databricks adapter, using a SQL warehouse for the SQL that dbt generates.