Skip to main content

dbt Supported Adapters Overview

A core principle of dbt is that it does not compute any data itself. dbt is essentially a highly intelligent SQL compiler and orchestrator. It writes the SQL, and then sends it to a data warehouse to do the actual heavy lifting.

But how does dbt know how to talk to Snowflake? How does it know how to talk to Google BigQuery? SQL dialects are different across platforms (e.g., Snowflake uses TIMESTAMP_NTZ, while PostgreSQL just uses TIMESTAMP).

dbt solves this using Adapters.

Topics Covered​

  • What is an Adapter?
  • Official (Core) Adapters
  • Community Adapters
  • How Adapters translate SQL

1. What is an Adapter?​

An adapter is a Python plugin that teaches dbt how to communicate with a specific database system.

When you install dbt Core on your laptop, you do not just run pip install dbt. You must specify the adapter you want to use.

For example, if your company in Bengaluru uses Snowflake, you install:

pip install dbt-snowflake

(This automatically installs dbt-core and the Snowflake plugin).

If you later move to a startup in Delhi that uses Google Cloud, you would install dbt-bigquery.

2. Official (Core) Adapters​

dbt Labs (the company behind dbt) officially maintains and supports plugins for the most popular modern cloud data warehouses. These are guaranteed to have full feature support and rigorous security testing.

The "Big Four" official cloud adapters are:

  1. Snowflake (dbt-snowflake)
  2. Google BigQuery (dbt-bigquery)
  3. Amazon Redshift (dbt-redshift)
  4. Databricks (dbt-databricks)

There is also a 5th officially supported adapter: 5. PostgreSQL (dbt-postgres) - This is primarily used for local development, tutorials, and very small analytical workloads.

3. Community Adapters​

What if your company uses a different database, like Microsoft SQL Server or Amazon Athena?

Because dbt is open-source, the global community of Data Engineers has built dozens of custom adapters. While these are not officially guaranteed by dbt Labs, many of them are heavily maintained by massive corporations.

Popular Community Adapters include:

  • dbt-sqlserver (Maintained by the community for MS SQL)
  • dbt-athena (Maintained by the open-source community for AWS Athena)
  • dbt-trino / dbt-starburst
  • dbt-spark
  • dbt-duckdb (Exploding in popularity for local, blazing-fast analytics!)

4. How Adapters Translate SQL​

The real magic of an adapter is not just the network connection. The adapter actually translates macros into the correct SQL dialect.

Let's say Priya wants to use a dbt macro called {{ dbt.date_trunc('month', 'order_date') }} in her SQL code.

If Priya is using the dbt-snowflake adapter, dbt compiles that into:

DATE_TRUNC('month', order_date)

If Priya switches to a company using the dbt-bigquery adapter, dbt compiles the exact same Jinja code into:

DATE_TRUNC(order_date, MONTH)

(Notice how the arguments are swapped? BigQuery expects the date first, Snowflake expects the part first).

Because of Adapters, Priya does not have to learn a brand new SQL dialect. She writes standard dbt Jinja, and the Adapter translates it perfectly for the target database!

Summary​

  • dbt connects to databases using plugins called Adapters.
  • The primary Official Adapters are Snowflake, BigQuery, Redshift, and Databricks.
  • There is a massive ecosystem of Community Adapters for tools like Athena, SQL Server, and DuckDB.
  • Adapters don't just connect to the database; they automatically translate dbt macros into the correct, database-specific SQL dialect.