Hacker News new | ask | show | jobs
by massaman_yams 3155 days ago
This covers the distinction a bit better. https://www.slideshare.net/HadoopSummit/the-columnar-era-lev...
2 comments
tedmiston 3155 days ago
The agenda slide says Kudu is mutable on disk while Parquet is immutable on disk.
link
jetblackio 3155 days ago
Right on, this is perfect. Thanks!
link
jacquesnadeau 3155 days ago
One quick note to make on this. Kudu is a storage implementation, (similar to Parquet in some ways). Arrow isn't about persistence and is actually built to be complementary to both Kudu and Parquet.

Also note: Kudu is a distributed process. Arrow and Parquet are libraries that can be embedded into your existing applications.

link