This test was done on a small dev laptop with 16GB of RAM, scanning a 500M row (record) 23GB Parquet file. DuckDB proved to be 5x faster.
A bit of an obvious one - small data tech is faster at small data. It serves more of a lower bound reminder of what "small data" is nowadays.
The article rightly starts with:
> Processing power on laptops has increased dramatically over the last twenty years. This allows single laptops to accomplish what we needed multi-node Spark clusters to do ten years ago.
