Apache Arrow
Hacker ApprovedThe most recommended of 3 columnar file formats
People recommend it as a shared, language-independent memory layout that lets tools pass data without copying; the most common caveat is that it has many bugs.
People recommend it as a shared, language-independent memory layout that lets tools pass data without copying; the most common caveat is that it has many bugs.
21 people gave an opinion
- recommend it
- 20
- criticize it
- 1
- approval share
- 95%
Recommended in 8 of 13 months
- November 2025: not recommended
- December 2025: not recommended
- January 2026: recommended
- February 2026: recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: not recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it carries the badge
- 20 people recommend itNeeds 10 or more, met.
- 20 recommenders for every criticNeeds 2 to 1 or better, met.
- Recommended in 8 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Shared memory layout across languages
“Protocols and IR (Intermediate Representation) have layed and continue to enable interoperability and composability of data tools (see Apache Arrow, Substrait, Catalog).”
Comment from 20 Mar 2026 on Hacker News “It's a standard memory layout available across applications and even across languages, which can be really convenient.”
Comment from 12 Feb 2026 in “Apache Arrow is 10 years old” “A standardized layout like Arrow means that the data is not tied to the layout/padding of a particular language”
Comment from 23 Jan 2026 in “Replacing Protobuf with Rust to go 5 times faster” - Comment from 12 Feb 2026 in “Apache Arrow is 10 years old”
- Comment from 12 Feb 2026 in “Apache Arrow is 10 years old”
Passes data between tools without copying
“The USP of Arrow ist that you can pass bits of memory between Polars, Datafusion, DuckDB, etc. without copying.”
Comment from 12 Feb 2026 in “Apache Arrow is 10 years old” “Apache arrow is great here, basically the reason we wrote the initial js tier is for easier shuttling”
Comment from 22 Jan 2026 in “Show HN: ChartGPU – WebGPU-powered charting library (1M points at 60fps)” Fast serialization and performance
“excellent performance for the actual serialization and deserialization part”
Comment from 7 Mar 2026 in “Nobody ever got fired for using a struct” “It worked flawlessly and 10-20x'd my throughput”
Comment from 12 Feb 2026 in “Apache Arrow is 10 years old” “The performance is so good.”
Comment from 12 Feb 2026 in “Apache Arrow is 10 years old” Usable C interface for bindings
“Arrow CPP for example has very portable builds and the C interface is very usable for building bindings.”
Comment from 19 Jun 2026 in “DuckDB Internals: Why Is DuckDB Fast? (Part 1)” “DataFusion and Arrow are AI-native, allowing for Python-Rust interoperability and speed”
Comment from 5 Oct 2026 in “Sail runs Doom with a single SQL query”
Other columnar file formats people recommend
The two most recommended other columnar file formats, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| Apache ArrowThis page | 20 | 1 | 95% | 8 | — | Approved |
| Vortex file format | 13 | 3 | 81% | 8 | — | Approved |
| Feather file format | 10 | 1 | 91% | 7 | — | Approved |
What people said
All 29 comments that mention it, newest first. Each links to the original comment.
“DataFusion and Arrow are AI-native, allowing for Python-Rust interoperability and speed”
“Using arrow is cool, but grpc seems like a more natural fit than arrow IPC”
“Arrow CPP for example has very portable builds and the C interface is very usable for building bindings.”
“Apache Arrow, that GOAT of data engineering”
“I was happily using and writing about Arrow as a data processing tool.”
“The Apache Arrow array format docs are a great read”
“there really aren't any good pure Swift libraries for Arrow or Parquet”
“Protocols and IR (Intermediate Representation) have layed and continue to enable interoperability and composability of data tools (see Apache Arrow, Substrait, Catalog).”
“excellent performance for the actual serialization and deserialization part”
“the right tool might be Parquet or Arrow or protobuf or Cap’n Proto”
“Arrow is also directly usable as the application memory model.”
“Arrow + GPU dataframes remain gifts that keep giving”
“they use signed integer types for buffer offsets and the like”
“It's a standard memory layout available across applications and even across languages, which can be really convenient.”
“there are just way too many bugs in Arrow”
“It worked flawlessly and 10-20x'd my throughput”
“It's efficient, complete and does not have "infinite precision decimals".”
“The performance is so good.”
“With common interchange formats, the benefits of these optimizations are shared across stacks.”
“a good chunk of polars' success can be attributed to Arrow which McKinney co-created”
“it's really a language-independent format so that you don't need to change it for say, a dataframe or R”
“The USP of Arrow ist that you can pass bits of memory between Polars, Datafusion, DuckDB, etc. without copying.”
“the impact it would have on the date ecosystem”
“try to enforce schema (eg, arrow) to keep typed & predictable”
“it is only the C Data Interface that can be FFI'ed, which is pretty low level”
“A standardized layout like Arrow means that the data is not tied to the layout/padding of a particular language”
“Apache arrow is great here, basically the reason we wrote the initial js tier is for easier shuttling”
“I think Apache Arrow has achieved exactly that”
“I really prefer Apache's approach to analytics where it's possible to pick and choose the parts you need”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.