No. The DuckDB Playground runs DuckDB-Wasm entirely inside your browser tab. Any file you add is read locally and never sent to a server.
DuckDB can query CSV, TSV, JSON/NDJSON, Arrow and Parquet files directly with functions like read_csv, read_json and read_parquet, or simply by selecting FROM the file name.
DuckDB is column-oriented and built for analytical (OLAP) queries — aggregations, joins and scans over large tables — whereas SQLite is row-oriented and tuned for transactional (OLTP) workloads.
DuckDB-Wasm is single-threaded by default and bounded by the memory available to your browser tab. It is great for exploration and files up to a few hundred MB; for very large datasets, use native DuckDB.
Yes. The Share button encodes the SQL in the URL so you can send it to someone. The query is shared, but uploaded data is not — the recipient runs it against their own data.
