dbt is an open-source data transformation framework that lets data analysts and engineers transform their data using the same practices software engineers use to build applications, and it is aimed at analytics teams that model data inside a warehouse.

What it is

dbt lives in the ELT and business intelligence space. Analysts write select statements, which dbt calls models, and dbt handles turning those statements into tables and views in a data warehouse. Models frequently build on top of one another, and dbt makes it easy to manage the relationships between them, to visualize those relationships as a directed acyclic graph, and to assure the quality of transformations through testing. The repository carries topics for analytics, business-intelligence, data-modeling, elt, dbt-viewpoint, pypa and slack, and the project documents its viewpoint alongside the introduction. The main branch now holds the Apache 2.0 source code of dbt v2.0, a ground-up rewrite of dbt in Rust, while the v1 Python implementation of the framework continues on the 1.latest branch.

The problem dbt solves is the gap between ad hoc SQL and maintainable data models: instead of tracking by hand which query must run before which, a team expresses a model as a select statement and lets dbt resolve the ordering through model references and check the result through tests. In the v2.0 line the rewrite also replaces the Python runtime and the dependency management that the v1 implementation required, because dbt v2.0 is distributed as a single self-contained binary. Analytics artifacts shift to Parquet, which can be readily queried, joined and analyzed, while the JSON artifacts such as manifest.json continue to be produced for backwards compatibility.

Key capabilities

• Models written as select statements, which dbt materializes as tables and views in a data warehouse.
• Relationship management between models through references, with documentation and DAG visualization of those relationships.
• Testing of transformations to assure their quality.
• A Rust rewrite in dbt v2.0 that parses, compiles and runs projects in a fraction of the time of v1, especially on the largest projects.
• A tightly defined language specification that enforces correctness at parse time.
• Parquet artifacts that can be queried, joined and analyzed, with JSON artifacts such as manifest.json retained for backwards compatibility.
• A single self-contained binary with no Python runtime or dependency management required, plus a revamped local documentation experience powered by the new artifacts.

Who uses it and how

• Analytics teams where analysts and engineers work in the same project: analysts write select statements while dbt handles the warehouse objects.
• Teams running large dbt projects, which is the scale the v2.0 parse and compile improvements target.
• Multi-person teams that want an enhanced collaboration experience through the hosted dbt platform.
• Practitioners who work through the dbt Community Slack and the dbt Community Discourse.
• Developers on macOS (x86-64 and ARM), Linux (x86-64 and ARM) or Windows (x86-64), since dbt v2.0 and its drivers are compiled per operating system and architecture.

Getting started

Install dbt following the installation documentation at docs.getdbt.com/docs/local/install-dbt?version=2, then read the introduction and the viewpoint. A hosted dbt platform is also offered for teams that want a managed collaboration experience.

How it compares

No comparable or paid products are named among the facts provided for this entry, so dbt stands alone in this registry. There is no basis in the available facts for a feature, licence or cost comparison against other tools.

When to use it — and when not to

dbt v2.0 is in beta, and its behavior, APIs and on-disk formats may change before the stable release, so teams that need frozen artifact formats or stable APIs should stay on v1 or wait. Anyone who wants the v1 Python implementation must switch to the 1.latest branch, and Windows on ARM is not yet supported. Note also the licence split: the source code in the repository is Apache 2.0, but dbt itself is a distribution of that repository with dbt-specific customizations released under a dbt product license, so self-hosters should confirm which terms apply to what they deploy.