# Apache Parquet

40 Hacker News commenters recommended Apache Parquet and 17 criticized it, November 2025 to November 2026. It was recommended in 12 of 13 months. It meets the Hacker Approved badge rule.

**Verdict:** Widely supported columnar file format that compresses well and lets tools read only needed columns. Critics say it is hard to append to and has heavy dependencies.

## Why people recommend it

- Columnar layout compresses well (8 comments): "I can get the zstd parquet to be up to 10x (or more) smaller than an equivalent Postgres table" ([source](https://news.ycombinator.com/item?id=48597866)) "with their built-in columnar compression will result in more performance AND space savings" ([source](https://news.ycombinator.com/item?id=49601697)) "You’ll see vastly smaller file sizes (even more so if you use zstd compression)" ([source](https://news.ycombinator.com/item?id=48596302))
- Reads only the columns needed (2 comments): "not only do you get seekable data, you can fetch only the columns you want too" ([source](https://news.ycombinator.com/item?id=46441135)) "DuckDB can do all manner of nifty optimizations while reading its own native format or Parquet" ([source](https://news.ycombinator.com/item?id=48648859))
- Strong support across tools (4 comments): "parquet has fantastic support and is still a pretty good choice" ([source](https://news.ycombinator.com/item?id=47000336)) "Parquet, Iceberg, and other open formats seem good" ([source](https://news.ycombinator.com/item?id=46447169)) "I think parquet and arrow are great formats" ([source](https://news.ycombinator.com/item?id=47271135))
- Good for analytics workloads (2 comments): "if you are just doing pure analytics and nothing else, Parquet will probably continue to do the job for you just fine" ([source](https://news.ycombinator.com/item?id=48650074)) "Parquet might win for table scans depending on the underlying storage medium" ([source](https://news.ycombinator.com/item?id=47289906))

## What people warn about

- Hard to append or update (4 comments): "Creating the Parquet data is a pain since it's not append-able" ([source](https://news.ycombinator.com/item?id=46993030)) "but with parquet you need large row group batches. Worst case (low log volumes and a time-based flush) you’d end up with loads of tiny row groups." ([source](https://news.ycombinator.com/item?id=49292472)) "you really want to avoid re-writing all 122880 rows of a row group every time a single insert happens" ([source](https://news.ycombinator.com/item?id=49292143))
- Heavy dependency footprint (3 comments): "The heavy dependency footprint of parquet-java was the main driver for kicking off this project." ([source](https://news.ycombinator.com/item?id=47206905)) "extremely unpleasant to use with its massive fan-out of dependencies" ([source](https://news.ycombinator.com/item?id=47206321)) "Pulling in Parquet and all of its dependencies is utter overkill." ([source](https://news.ycombinator.com/item?id=47245770))
- Does not fix encoding problems (2 comments): "while Parquet fixes the "delimiter problem", it doesn't fix the "encoding problem"" ([source](https://news.ycombinator.com/item?id=47246091)) "A big problem with parquet, which this aims to replace, is that it's hard to add new encodings" ([source](https://news.ycombinator.com/item?id=48648621))
- Not the best at compression (2 comments): "Parquet is also 13 years old and not the GOAT at compression" ([source](https://news.ycombinator.com/item?id=49991563)) "Plus isn't the least wasteful format" ([source](https://news.ycombinator.com/item?id=47429123))

## Alternatives hackers recommend

- [Safetensors](https://hackerapproved.com/p/safetensors.md): 7 recommend
- [JSON Canvas](https://hackerapproved.com/p/json-canvas.md): 7 recommend
- [F3 file format](https://hackerapproved.com/p/f3.md): 6 recommend

## Cite as

Hacker Approved, "Apache Parquet", https://hackerapproved.com/p/parquet/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).