Overview

Tangram OS SQL is a multiple engine service implemented using PostgreSQL wire protocol. It enables seamless data exploration, analytics, and transformation across multiple data platforms from a single interface.

Tangram OS SQL Engine supports integration with a variety of query engines, including: DuckDB, ClickHouse, Databricks, and more...

Key Features

By abstracting the underlying engines with a open protocol, Tangram OS allows users to:

Seamlessly Query Open Table Formats Across Engines

Tangram OS SQL natively integrates with open table formats like Apache Iceberg and Delta Lake, eliminating the need for complex engine-specific setups. You can query these tables as if they were native to the engine. For example:

  -- create an Iceberg table
  create table iceberg.test.nyc_taxi(time timestamp, district string, miles double);

  -- insert some data 
  insert into iceberg.test.nyc_taxi(time, district, miles) 
      values('2025-01-01 17:39:09.439', 'Brooklyn', 12.6);

  -- use DuckDB to query a Iceberg table
  use engine duckdb;
  select sum(miles) as total_distance from iceberg.test.nyc_taxi group by district;

  -- use ClickHouse to query the same table
  use engine clickhouse;
  select sum(miles) as total_distance from iceberg.test.nyc_taxi group by district;

Unify data access control with Tangram IAM service

As the following example, Tangram SQL will honor the access policy defined via the example role, will allow users with that role to have read-only access to Iceberg tables in workspace demo and test Iceberg database.
Maintain engine-specific optimizations
While providing a unified interface, Tangram OS still enables users to utilize engine-specific SQL dialects and capabilities as needed.

Key Features​

Key Features