4 ms·
Frankly I don't think ZFS code complexity should be touted as an advantage.
by jmnicolas 9y ago
Frankly I don't think ZFS code complexity should be touted as an advantage.
- ianai 9y agoNot in isolation. But everything I’ve read about the implementation says many engineering hours were poured into it.
- late2part 9y agosun built it, oracle bought it, freebsd popularized it, linux has it.
- pizza234 9y agoThe complexity comparison itself is not correct - new generation "file systems" are actually both volume managers and file systems (which also implies other features that live at the intersection).
- masklinn 9y agoThe article properly points this out too.
- kev009 9y agoThe graph seemed to be of estimated development years. If you accumulate an in place FS, dm, mdraid, LVM and all assorted userland tools I think that ZFS is probably less code complexity. There is some ideological complexity, namely the ARC (in implementation, not in algorithm) is a pretty big wart. It works well when properly dressed, but it is pretty hard to make work well with most OS kernel allocators and memory pressure.
- wazoox 9y agoAnd I'm pretty sure it's complete bullshit. Ext4 and XFS have both dedicated full-time developers working on them for many years. Unless they simply use a simple SLOC rule of the thumb, which is pretty much invalid for such large, long-running projects with highly optimized code.