Hacker News Reader

Comment by justincormack | Hacker News Reader

"The CDDL prevents ZFS from being included/heavily integrated into Linux, the BSDs, OS X and Windows" - no, it is included and integrated into FreeBSD with no issues, NetBSD has an old version that needs updating, OpenBSD thinks it is too complex. OSX could use it (they could get a commercial license anyway), but seemed to decide not to, as could Windows.

View on HN

I think the actual difficulty is patents. From what I remember, once Oracle bought Sun, they decided to go after netapp for infringing on ZFS. As it turned out, Oracle was infringing on NetApp's patent but NetApp had let it slide because until Oracle started being a dick, it was mutually beneficial. That is why Oracle also started BTRFS and is putting it's development behind that instead.

I'm not really sure where that leaves other commercial projects (especially OSX which is badly in need of a functional file system).

It was really the other way around. NetApp had Sun in court over ZFS infringing on WAFL patents[1] well before the buyout.

I certainly don't know all the ways people use their Macs, but it seems like ZFS would not get a lot of use in OS X unless it was the only option:

- It doesn't offer a whole lot over traditional filesystems in a single-device context; since Apple has basically abandoned server/enterprise, I would wager the vast majority of new OS X systems are single-drive.

- At the time Apple was supposedly making a decision about this, I am not sure if ZFS handled 4k alignment (ashift); this is very important now that a lot of new Macs are shipping with SSDs.

- You're still, even w/ 10.10, discouraged from using case-sensitive HFS+; at least one application[2] won't work on a case-sensitive filesystem on OS X.

The above reasons probably apply to Windows as well. Also, in general, it seems like both vendors want to be the sole source for all of their core features.

[1] http://www.zdnet.com/article/netapp-claims-suns-zfs-violates...

[2] https://support.steampowered.com/kb_article.php?ref=8601-RYP...

Thanks for the links, it had been a while so I was a little foggy on that one. In the end, I don't think you'd want to back the project for that reason.

>It doesn't offer a whole lot over traditional filesystems in a single-device context; since Apple has basically abandoned server/enterprise, I would wager the vast majority of new OS X systems are single-drive.

Not really, but of all the major vendor file-systems, HFS+ is probably the weakest link. An update to a filesystem that is well maintained for a freeBSD base would probably be an upgrade. Also, even with a single disk, zfs would have the benifits of good filsystem level compression which can boost read write/speeds by 2 or 3 times. Especially on a no server box where you are likely to have idle cores.

There are some features from an os level of using COW if you want to give people restore points or built-in os VCS for files. Also checksums (ie, this file looks to be corrupted, would you like to go back to yesterday?).

So there's some cool usage for these file systems even on a single disk but again a corporation probably doesn't want to touch it with a 10 foot stick.

From link [1] above it looks like Sun went after NetApp, after NetApp poked around, it turned out that Sun was the one in violation. I remembered that was the case but I was thinking that oracle had already bought them at that point

FTA:

>Sun approached NetApp about 18 months ago with claims the storage maker was violating its patents and seeking a licensing agreement, NetApp Chief Executive Dan Warmenhoven said in a statement.

>Several months into those discussions and following a review of the matter, NetApp made a discovery of its own, Warmenhoven said, concluding NetApp did not infringe the patents but that Sun infringed on NetApp's.

That's not at all what happened. (Disclaimer: I was at Sun at the time and was deposed in the case.) Sun didn't "go after NetApp" -- NetApp tried to buy some StorageTek patents via a third-party intermediary, and when they were rebuffed, they came after ZFS.[1] And, it should be said, NetApp didn't particularly care about Sun -- they cared about the fact that ZFS was open source. NetApp wanted Sun to "close" ZFS or otherwise "restrict its use"[2]. As for the case itself, it was moved back to California (NetApp had initiated it in East Texas, the patent troll capital of the universe) where it became a massive case, and was then slimmed down by order of the magistrate to three patents on the NetApp side and four patents on the Sun countersuit side. At the same time, thanks to a community outpouring of prior art, Sun was able to pursue invalidating the claims of the NetApp patents with the US Patent Office.[3] These efforts were wildly successful, and all three NetApp patents were rejected on all claims. Amazingly, the case wasn't thrown out at that point (though any damages would obviously be very limited), but every turn in the case had gone Sun's way.

Then, Oracle acquired Sun, and for reasons that haven't been disclosed, Oracle and NetApp dismissed their respective cases.[4] While I can't disclose the reasons behind this, I can say that both Oracle and NetApp would have jumped at the chance to cross-license ZFS and WAFL patents in a way that extended only to Oracle and not to CDDL licensees. (That is, prohibited open source ZFS.) Because the CDDL is airtight with respect to patents, such cross-licensing was impossible, and by dismissing their suits (instead of settling), the findings of fact from the trial essentially disappear -- which is enormously to NetApp's advantage. Point is: ZFS actually has about as much patent security as one can find in an open source system, as it has withstood a direct, full-frontal assault by attorneys seeking to find a way around its patent grants.

[1] https://web.archive.org/web/20070423001711/http://blogs.sun....

[2] https://web.archive.org/web/20080625023043/http://blogs.sun....

[3] http://www.theregister.co.uk/2008/10/07/sun_gets_netapp_pate...

[4] http://www.theregister.co.uk/2010/09/09/oracle_netapp_zfs_di...

Oh wow, thanks for the context and links.

ZFS has supported autodetecting 4k drives since day one, with the caveat that certain drives that were programmed to try to fool OSes that only supported 512 byte sectors would occasionally fool ZFS as well [1]. Additionally, ZFS can be set as case-insensitive, to match HFS+ behavior. Finally, I'd argue that ZFS is useful even to the one-drive user. The snapshot feature would make Time Machine backups even more transparent, compression would make things a bit smaller, and filesystem sending/receiving would make backing up to other devices a lot simpler. Modern filesystems just make sense, even for the regular user.

[1] http://blog.delphix.com/gwilson/2012/11/15/4k-sectors-and-zf...

Apple's Core Storage is a logical volume manager, and it has COW features, in particular when using its dmcache-like function to marry an SSD and HDD into a single logical volume. The could conceivably tack on checksums at this level (instead of parity for example). Core Storage volumes are the default since 10.10, in fact as long as you don't already have a Boot Camp'd dual-boot setup, the Yosemite install process converts the main partition to a Core Storage LV in-place.

Reply on news.ycombinator.com