Skip to main content

DuckDB

DuckDB is an open source, in-process, analytical database. Its can natively read several of output formats that Fused HTTP Endpoints return.

1. Generate a signed URL for a UDF​

First create a UDF and generate an HTTP endpoint.

2. Install and load httpfs​

To load Parquet files from remote endpoints from within DuckDB, you can install the httpfs extension.

INSTALL httpfs;
LOAD httpfs;

3. Query using read_parquet​

Now you can make a query using the UDF URL, with the dtype_out_vector set to parquet:

SELECT *
FROM read_parquet('https://www.fused.io/server/v1/realtime-shared/221aa65f3d96f1a320ed0f4eea0d320724c0ddc0c75cbf70df711def11e2ecc5/run/file?dtype_out_vector=parquet');

You can pass parameters into the URL from the query:

SELECT *
FROM read_parquet('https://www.fused.io/server/v1/realtime-shared/221aa65f3d96f1a320ed0f4eea0d320724c0ddc0c75cbf70df711def11e2ecc5/run/file?dtype_out_vector=parquet&resolution=13');