DuckDB is a genuine time-saver for the small jobs — the ones where you have a file, a question and no appetite for setting up a database first. It is bundled with SQL DATA LENS, so there is nothing to install and no driver to download.
Query a file where it lies
Most of the time the question is something like “what is actually in this export?”. With DuckDB you can answer it against a CSV or Parquet file directly:
SELECT * FROM 'C:\path\to\my\datafile.csv';Auto-detection works out the delimiter, the quoting and the column types, so a first look costs one statement. The same syntax works for Parquet and JSON files.
Creating an in-memory connection
To get started, right-click in the connection tree and create a new in-memory DuckDB connection. There is a keyboard shortcut for it as well, so the whole thing is a couple of seconds’ work.
Then open a query window on that connection and start working on your files.
Because the connection is in memory, nothing is written to disk and closing it cleans up after itself. That is usually what you want for ad-hoc analysis; create a file-backed DuckDB connection instead when you want the results to survive.
Faster still: drag and drop
The quickest route skips the typing altogether. Drag a file onto the DuckDB connection in the tree, and SQL DATA LENS writes the statements for you — a SELECT, plus DESCRIBE and SUMMARIZE so you get the column types and the per-column distribution in the same pass.
SUMMARIZE in particular is worth the habit: min, max, distinct count and null count for every column, which is often all the profiling a file needs before you decide whether to load it anywhere.
Where it fits
DuckDB does not replace your IRIS connection — it sits next to it. See beyond IRIS for the other engines that are reachable from the same window, and the SQL editor for the editor features that apply to DuckDB queries too. The bundled build is on the download page.