RAWQL / LOCAL SQL WITH DUCKDB

Query Parquet in your browser

Inspect a Parquet export with DuckDB SQL without setting up a database server or uploading its contents to RawQL.

RawQL is in invite-only beta. These guides and sample files are public. You need beta access to run the examples in the editor.

Import a small file first

Download orders.parquet. It contains six fictional orders with the same values as the CSV guide. Open Databases → Data files and import it. RawQL registers the view orders_parquet.

Inspect the schema

Parquet stores column types alongside the data. Check those types before writing a calculation:

DESCRIBE orders_parquet;

The sample has integer order and customer IDs, text status and a decimal amount. For your own exports, also check timestamps, nullability and nested columns in the catalog.

Select only the columns you need

SELECT order_id, amount
FROM orders_parquet
WHERE status = 'paid' AND amount >= 100
ORDER BY amount DESC;

DuckDB can push column selection and eligible filters into its Parquet reader. That can reduce work during a query. RawQL still loads the local file into browser memory, so this does not remove the import size limit.

Summarize the whole export

SELECT status, COUNT(*) AS orders, SUM(amount) AS amount
FROM orders_parquet
GROUP BY status
ORDER BY status;

Export and preserve your work

Export the query result when you need a smaller extract. Keep the SQL in a worksheet. A view stores a query definition rather than a copy of the source data; keep its source file loaded, or download a workspace snapshot before ending the session.

Limits and common mistakes

The beta accepts up to three loaded files and 100 MB per file. Compressed file size is not the same as memory required for a query. Start with a small selection and avoid expanding every nested value at once. A LIMIT caps returned rows but does not guarantee that an aggregation reads less input.

This workflow is for local files. It does not promise remote object-storage access or server-scale memory. See DuckDB Parquet documentation for reader behavior and supported SQL options.