Live data from Hacker News

Native port of ZFS for Linux

github.com

21–30 of 32 posts

Re: Native port of ZFS for Linux

#21

They claim that they're legally free and clear, but a lot of the kernel devs are of the view that kernel modules are derived works of the kernel and therefore have to be GPL.

Given all the binary drivers available for Linux, those people are clearly toothless.

Re: Native port of ZFS for Linux

#22
post #5

Did the ZFS patents ever get sorted out? The US government needn't worry about this, they have an exemption, but the rest of the world should care. Sun has a truckload of patents on ZFS that, last I'd heard, they were not granting free to ZFS users, and NetApp thought they had a patent on it as well, at least enough to sue Sun.

I don't think the rest of the world cares that much. US software patents don't mean much to anyone outside the country...

Re: Native port of ZFS for Linux

#23
post #5

Did the ZFS patents ever get sorted out? The US government needn't worry about this, they have an exemption, but the rest of the world should care. Sun has a truckload of patents on ZFS that, last I'd heard, they were not granting free to ZFS users, and NetApp thought they had a patent on it as well, at least enough to sue Sun.

I don't think the rest of the world cares that much. US software patents don't mean much to anyone outside the country...

Yeah I doubt RIM would agree with you. That is just one example.

Re: Native port of ZFS for Linux

#24
post #19
post #14

I'd like to know how this could be legally redistributable, given the license conundrum CDDL and GPLv2 make. Any idea?

ZFS is redistributable and Linux is redistributable; you just can't redistribute the combination. So you'll have to download and install ZFS separately.

Mmmh, interesting. Apparently they did this with production in mind, so it must be usable. I'll give it a try...

Re: Native port of ZFS for Linux

#25

Would love to see ZFS (or at least something better than HFS) on OS X. Anyone has info on that? Anything in the pipeline?

Apple had one point had read/write support for ZFS, but it was pulled because Sun at the time was not able to provide certain guarantees. There was an email by one of the main guys in ZFS that said something to that effect. As such ZFS on Mac OS X is a dead project.

I think PureDarwin folks are working on this. http://www.puredarwin.org/downloads/xmas

Re: Native port of ZFS for Linux

#26
post #19
post #14

I'd like to know how this could be legally redistributable, given the license conundrum CDDL and GPLv2 make. Any idea?

ZFS is redistributable and Linux is redistributable; you just can't redistribute the combination. So you'll have to download and install ZFS separately.

Thank you. I was looking for some discussion here, but everyone was just spouting off about patents, and it being "illegal", which isn't the case. OpenAFS has a similar license incompatibility, yet it's still very widely used.

Re: Native port of ZFS for Linux

#27

They claim that they're legally free and clear, but a lot of the kernel devs are of the view that kernel modules are derived works of the kernel and therefore have to be GPL.

They may or may not be "derived works" (whatever those are), but they aren't "derivative works", which is what matters as far as copyright law is concerned.

GPL people have a tendency to forget that GPL is a copyright license, and so if someone is doing something that doesn't require copyright permission GPL is irrelevant.

Re: Native port of ZFS for Linux

#28
post #15
post #12

Seems bit old, originating from 2008? Any notable progression lately?

It went nowhere. Linux distros and the mainline kernel are using btrfs, which has a similar layer-spanning design as zfs but is btree based like Reiser (which should make it faster). Most features are done and you can install with btrfs out of the box today in the RHEL 6 or Centos 6 betas. http://en.wikipedia.org/wiki/Btrfs

ZFS has the following that BTRFS doesn't * Deduplication - Huge in any virtualization environment * Cache layer - Required for high IOPS * fsck - BTRFS cannot fix itself * raid - BTRFS raid currently cannot repair itself and current git of the latest linux kernel still has these isses as well as space issues. I wouldn't trust BTRFS to store /dev/random

Re: Native port of ZFS for Linux

#29
post #23

Earlier quoted context omitted.

I don't think the rest of the world cares that much. US software patents don't mean much to anyone outside the country...

Yeah I doubt RIM would agree with you. That is just one example.

I don't get this reference. Links?

Re: Native port of ZFS for Linux

#30
post #15

Earlier quoted context omitted.

It went nowhere. Linux distros and the mainline kernel are using btrfs, which has a similar layer-spanning design as zfs but is btree based like Reiser (which should make it faster). Most features are done and you can install with btrfs out of the box today in the RHEL 6 or Centos 6 betas. http://en.wikipedia.org/wiki/Btrfs

ZFS has the following that BTRFS doesn't * Deduplication - Huge in any virtualization environment * Cache layer - Required for high IOPS * fsck - BTRFS cannot fix itself * raid - BTRFS raid currently cannot repair itself and current git of the latest linux kernel still has these isses as well as space issues. I wouldn't trust BTRFS to store /dev/random

Good, because it's an experimental filesystem. Don't write it off completely though, because all of those (except for "cache layer" -- wtf does that even mean?) are planned.
Post reply on HN