loremfile

Sample Parquet files for testing data pipelines — free, CC0

Apache Parquet is the columnar format of data lakes and analytics engines. It stores each column separately and compresses it, so queries read only the columns they need.

parquet/people-1000.parquet and parquet/people-100k.parquet hold 1,000 and 100,000 rows of the shared synthetic people dataset, snappy-compressed, each in a single row group. They contain the same records as the CSV, JSON, Arrow and Avro people files, so a pipeline can be checked for identical results whichever format it reads.

The two sizes test different things. The small file is quick to inspect with any Parquet reader; the large one is big enough to measure column pruning, filtering and memory use, and to test upload paths that stream rather than buffer. A single row group keeps the layout simple, so row-group statistics are easy to reason about.

Use them with DuckDB, pandas, Spark or cloud query services. Related formats: Arrow, Avro, CSV and SQLite.

2 files. The same list as JSON: https://loremfile.dev/parquet/index.json

Parquet files
FileSizePropertiesDescription
people-1000.parquet 114.5 KB 1,000 rows 1,000 rows of the shared people dataset as Parquet, snappy-compressed, in a single row group. The same rows as every other people fixture.
Snippets 
curl -O https://loremfile.dev/parquet/people-1000.parquet
<a href="https://loremfile.dev/parquet/people-1000.parquet">download</a>
people-100k.parquet 9.8 MB 100,000 rows 100,000 rows of the shared people dataset as Parquet, snappy-compressed, in a single row group. The same rows as every other people fixture.
Snippets 
curl -O https://loremfile.dev/parquet/people-100k.parquet
<a href="https://loremfile.dev/parquet/people-100k.parquet">download</a>

File names follow the naming grammar. Verify a download against sha256sums.txt.