Linux Torvalds on ZFS and the Linux kernel

Instead of posting a link directly to the Linus Torvalds interview,
I’ll post a link to the Ars Technical criticism… Not because I’m taking any side but because the article contains some opinion and statement that relates to BTRFS, and can help some come to know issues related to the main alternative to BTRFS

In this article, the author dings Linus Torvalds for opining on a topic he possibly knows very little and offers his opinion on why the main alternative to ZFS isn’t ready for Prime Time…

There’s only one other widely available filesystem that even takes a respectable stab at providing most of those features, and that’s btrfs—which was not available for the first several years of ZFS’ general availability. In fact, btrfs still isn’t really stable enough for production use, unless you nerf all the features that make it interesting in the first place.

I don’t know that I’ve ever heard a BTRFS maintainer oversell BTRFS or for that matter state a claim that would need to be tested, but I think that everyone in these openSUSE Forums can note how the number of BTRFS related posts needing help has dropped off considerably over the years. It still has a ways to go, but if one were to skim through the Ars Technica viewer comments, one could notice a very large number of systems with 4 disk arrays which AFAIK should be quite suitable for using BTRFS in place of ZFS.

I’m not going to criticize anyone for using ZFS, but will only observe how much more work is involved to set up and maintain what in the end in features approximates what can be set up almost automatically with next to zero effort in BTRFS. We who use BTRFS should still note how some features have been used and tested plenty(eg snapshots) and some used quite a bit less (RAID) and for many nothing can replace a record of longevity to establish a record of reliability. I’ve also never benchmarked BTRFS vs ZFS nor have had any practical reason to investigate, I’ve always felt that at least how I use openSUSE BTRFS has been “fast enough.”

But, the technical issues of BTRFS and ZFS actually were actually the side show in the latest ZFS dustup…
The central issue is that ZFS is not distributed in the kernel, the reason that’s the case and why that won’t change soon(legal issues surrounding Oracle’s possible patent and copyright claims).
That, and the change in the kernel this past year probably introduced a bit of risk and uncertainty into using ZFS, perfectly within established policy but making ZFS Users mad as heck.


Well according to Linus ZFS was not maintained anymore. Also the license is not compatible with GNU’s GPL. Remember the law suit that lasted for years by some company that claims they own UNIX copyright? I think that’s the thing that Linus is trying to avoid and of course Linus is being Linus as always… :wink:

Published criticisms of Torvald’s statement that ZFS “is not maintained” are not sure what he was referring to.

The original ZFS code created and maintained by Sun before sold to Oracle is certainly not maintained anymore…
But when the Sun patent rights were transferred to Oracle, there was apparently a revolt and exodus similar to what happened with OpenOffice and MySQL, and OpenZFS was forked and from then diverged. Oracle took the code “closed” at that point and created Oracle ZFS.

So the way it has been described is that today Torvalds might have been referring to any of the three variations of ZFS…

  • The original Sun ZFS, which certainly isn’t maintained anymore but it’s the one with the clearest legal status as GPL’d.
  • The Oracle ZFS which is closed source so can never be considered to be included into the mainline Linux kernel.
  • OpenZFS which is licensed GPL, but the GPL might be defective. Because there is someone (Oracle) who is known to be litigious about even borderline patent claims, Torvalds is saying the Linux kernel doesn’t want to even risk any patent claim. Torvalds demands a clear statement from Oracle renouncing all claims even to OpenZFS and Oracle won’t provide it. It might be useful to compare to OpenOffice/LibreOffice and MySQL/MariaDB, their histories mirror what happened to ZFS/OpenZFS. The big difference of course is only OpenZFS would like to be accepted into the kernel and AFAIK the others aren’t contributing code or want changes to be made in the kernel. So, similar to LibreOffice and MariaDB, OpenZFS might be accepted by organizations like SUSE which have strict FOSS standards but the Linux kernel seems to have a different legal opinion.

Whatever Torvalds is referring to about maintenance, OpenZFS is an active project.


Referring to the SCO claim that Linux violated its copyrights?–Linux_disputes

Now that’s an interesting question,
In 2007 Novell was victorious on behalf of all Linux…
But since then there have been a number of changes of ownership.
I wonder who owns those patents and copyrights now… SUSE? Microsoft? Someone else?


I’m not competent enough to intelligently discuss filesystem design, but I have used ZFS with FreeBSD and with Xubuntu. On Xubuntu, it sux. Period. Stick with EXT4 on Linux systems and use ZFS on *BSD systems, although OpenBSD appears to be only useful with UFS.