Hacker News Reader

The zstd estimator is not that great, so you should definitely keep the auto-detection of the file system on.

If you try to compress a completely incompressible file, it will still be extremely slow at higher compression levels: https://github.com/borgbase/vorta/issues/485#issuecomment-88...

View on HN

I can't recall if they actually implemented or just talked about it, but during the ZFS team meetings there was some discussion about using LZ4 as a "compression canary", to determine if it was worth attempting Zstandard. That is, first try compressing with LZ4, and if it did compress, do actual compression with Zstandard.

IIRC the argument went that LZ4 had a very quick bail-out routine, and quite fast compression, so the overall overhead would be quite small, at least for the higher Zstandard compression levels.

They implemented it; it landed in May: https://github.com/openzfs/zfs/commit/f375b23c

Quite the reverse actually, the btrfs estimator is way worse than the zstd one. By using the btrfs estimator, you leave a ton of compression ratio on the table, for no good reason.

Reply on news.ycombinator.com