Drop support for zstandard
Open
- Dominant language
- Python
- Stars
- 1.7k
- Forks
- 778
- Avg merge
- 2h 50m
- Merged PRs (30d)
- 3
Description
Following A/B tests con Coiled, zstandard seems to be, consistently, a poorer choice than lz4.
I do not see any reason why one would prefer it to lz4 andI think we should drop it in order to reduce our support surface.
A/B tests of zstandard vs. lz4, used for both network comms and spilling, exhibit a 15~80% slowdown in end-to-end runtime across the board:
XREF #2970
XREF #7403
Contributor guide
Assessment
This issue has not been assessed yet.