Native ZFS Is Coming To Linux Next Month 273
An anonymous reader writes "Phoronix is reporting that an Indian technology company has been porting the ZFS filesystem to Linux and will be releasing it next month as a native kernel module without a dependence on FUSE. 'In terms of how native ZFS for Linux is being handled by this Indian company, they are releasing their ported ZFS code under the Common Development & Distribution License and will not be attempting to go for mainline integration. Instead, this company will just be releasing their CDDL source-code as a build-able kernel module for users and ensuring it does not use any GPL-only symbols where there would be license conflicts. KQ Infotech also seems confident that Oracle will not attempt to take any legal action against them for this work.'"
Re:Freedom ain't free (Score:5, Insightful)
Sun used the CDDL just to make sure Linux never got ZFS. Even that move is not going to save solaris, only open sourcing it earlier would have done that. I say this as a linux user who likes solaris and thinks it will be a shame to see it die. Well I like it once the GNUtools are installed, the solaris versions sucked.
They are both quite open, how free they are some might argue about.
Good Article (Score:5, Insightful)
No, really. I had a bunch of questions going in, and they were all answered. This is rare enough to warrant a shout out to Michael Larabel.
I disagree with some of his subjective claims like x86_64 being a substantive limitation or ZFS on Linux remaining niche (I guess that depends on how you define the niche...) but he got the national lab project, the zpool version, the Oracle (nee Sun) patent problem. Kudos.
FreeBSD 9 is probably where ZFS will wind up finding a proper home, I'm guessing.
Re:Good Article (Score:5, Insightful)
How do you think it is not a substantive limitation?
My phone runs linux and is not x86 of any shape or register size, nor is my workstation, nor are many other machines I have running linux. This is just like people who think flash working only on x86 32bit linux is good enough.
If FreeBSD ever gets a good ZFS implementation expect lawsuits to commence.
Re:Open Source != Free Software (Score:4, Insightful)
I don't know if that's true. I know you probably can't redistribute the kernel with the CDDL bits but you can redistribute them separately (CDDL = Common Development and Distribution License). Then all you have to do is make sure that your software (or customer) installs the right bits and then you can get a pretty decent NAS box.
Besides the legal issues, I would love to see them tackle the technical issues. ZFS itself is very clean in code, very well documented and pretty simple once you get down to the wire. The issue (and selling point) is going to be performance and upkeep and for commercial implementations support. If the upkeep is going to be similar to BSD's implementation (several versions behind) or the performance as bad as FUSE, people are just going to stick to OpenSolaris (or one of it's commercially supported decendants like Nexenta).
Re:who cares?! (Score:3, Insightful)
ZFS has becoming vapor ware since apple announced snow kitty wasnt gunna support it.
I do not think that word means what you think it means.
Re:Freedom ain't free (Score:4, Insightful)
No, they are a company that exists to make money. Saving Solaris would make them more money. Very simple. Corporations do not hate like that, they only do what they must to maximize profit.
BSD is a fine license, it was created for a real purpose, not to just protect a doomed product.
Re:Open Source != Free Software (Score:4, Insightful)
This is both Open and Free, just not quite as free as Stallman would like.
CDDL licensed code can be freely distributed and modified, so long as it is compiled with a compatible license.
This is why BSD has no issues with including ZFS. The BSD license is less restrictive than the GPL.
Re:If it comes out and works well (Score:3, Insightful)
Not that I doubt you, but have any references to verify that claim please?
Re:If it comes out and works well (Score:2, Insightful)
If you think LVM + ext3 is better than NTFS sorry that speaks only to ignorance and possible Linux zealotry, not to knowledge. I don't care if you like MS, their file system is high class. It offers good performance and a very wide feature set, and they update it all the time. They don't rename it, but NTFS has been improved with each version of Windows.
No shame in not knowing about file systems, it is rather esoteric, but then please don't go shooting off at the mouth about how $your_chosen_platform has the bestest FS and everything else sucks.
Linux actually is a good deal behind in the FS game, which is one of the reasons there's so much interest in btrfs or ZFS. Linux could use a more modern file system for many tasks.
Finally Btrfs doesn't "Blow NTFS out of the water," because btrfs cannot be used on production systems. NTFS is used on basically every production Windows system, desktop or server, in the world. Btrfs is still under development. It may be a far superior file system when it comes out, but it isn't out yet. You can't set it up and use it and expect all your data to be ok.
There's a big difference between something under development and something in production. IE9 doesn't "Blow Firefox 3 out of the water," either. In theory it will, I've tried the test builds and its rendering is amazingly fast because of the hardware acceleration. However, it isn't a stable, release browser. It is just a test. How it does in realty when it is released, and equally important how FF is doing at the same time, remain to be seen. Making any claims of it at this point would be extremely premature.
Re:Freedom ain't free (Score:2, Insightful)
Free Software has a tendency to get stuck at "good enough" sometimes.
Not just free software. IE6 has the poster child of getting stuck at "good enough" until it was kicked out of place.
Re:If it comes out and works well (Score:3, Insightful)
So you are suggesting I can freeze IO to the machine, then run a snapshot command on NTFS?
I would be glad to hear it.
Re:Good Article (Score:3, Insightful)
Don't be intentionally dense. The majority of the market that can help these guys refine their code is fine with x86_64.
Re:If it comes out and works well (Score:3, Insightful)
Well when it comes to filesystems, I wouldn't want to run one on any production machine until the developers say they are stable. Bad FS driver can equal file system corruption and data loss.
The problem with testing early release stuff is that it is not necessarily representative of the final product. It can, of course, be too slow because of lack of optimization. However it can also be too fast. What do I mean by that? I mean perhaps during implementation, they discover that more extensive checks and processing of certain kinds is necessary to maintain data integrity. This ends up slowing things down.
I'm not saying that will happen, just saying it can, hence I don't like tests until the final version is there and everything is in it.
Also the other problem I have is that you don't know if something will ever reach completion. I've seen OSS projects that rocketed to a stable release in record time, I've seen them proceed slow and steady to a release, and I've seen them peter out and stagnate. You really don't know how it is going to be until it comes out.
I remember when at work we were trying to decide on an e-mail client to recommend to users (I work in an academic environment so we can't mandate things). None of us were that happy with Outlook for IMAP (though in 2010 it has gotten a lot better). I was using Thunderbird and found it to be acceptable. Since it was free and worked, I said lets do that. My boss wanted to wait for another one, a project he'd been watching for some time (can't remember the name). Was going to be awesome when it released... Ya well that day never came. They kept talking and talking, but no releasing of final code. Maybe some day it will be awesome, but right now Thunderbird is 100% more awesome since it is out, running and useful.
I have high hopes for btrfs because our large, inexpensive, non-backed up storage (our backed up storage is NetApp) at work is Linux and it could use something better than ext3. However I won't say "Yes this is a good FS," until it is a released product and I've seen how the final thing works.
Re:Good Article (Score:3, Insightful)
What's intimidating?
Being a hobbyist OSS developer and getting hit with a patent infringement lawsuit from a large corporation.
Re:If it comes out and works well (Score:2, Insightful)
You ready to try it on a live Oracle database server with a 10TB filesystem that averages at least say 5,000 DB transactions per second? :)
Re:Useless without dedupe (Score:1, Insightful)
1. Dedup is already part of ZFS - just not in an official release yet. You can get it in the dev builds of OpenSolaris. However, because Oracle has canceled OpenSolaris, you will not likely see an official dedup-enabled ZFS until Solaris 11.
My understanding, though, is that Sun has some storage solutions in production /w ZFS dedup that you can buy today.
Considering that no other open source file system offers this feature, it seems like a stupid criticism.
2. Those characters can be represented in UTF-8 just fine, and I just created a folder with that exact name. No problems.
Re:Freedom ain't free (Score:3, Insightful)
I'll be the first to say that ZFS has some shortcomings and limitations.
However, it's like Active Directory is to the workstation/server model of enterprise networking: it does everything. There is nothing else which comes close (or shell we say, 9/10ths of the way) to it in terms of it's "completeness" and feature set. Yes, it has some severe limitations ("Windows only") and shortcomings ("OMG it's a pain to troubleshoot"), with a fairly deep learning curve as well as a limited domain of applicability beyond the base subset (network administrators/storage techs).
But at the end of the day, they do things easily which most other products can't even do in such a complete fashion. Before such capability can be surpassed, it has to be met. AD and ZFS have been out now in more-or-less their current incarnation for close to 7 years, and only a bare subset of those features are elsewhere (and in less-than-ideal development status).
Frankly, ZFS is a (the) "next generation filesystem" for these reasons. It made assumptions (which are wrong), but do not make it lack utility. The management toolset is clean (very clean).
How, exactly, was being the conceptual "next generation" not the conceptual goal for the first 64 bit, CoW, RAID-built-in filesystem available? Seems to me that's why it's the fox everyone is chasing...
Re:Freedom ain't free (Score:4, Insightful)
Instead, btrfs, hammer, etc were developed -- much better, much cleaner file systems.
How can filesystems that don't exist in stable release form yet be "better" than ZFS?
ZFS is far ahead of btrfs, both in terms of stability, features, and usability. Btrfs doesn't have parity RAID, dedupe, or replication yet. These are critical features for large-scale systems. In short, it isn't even close to ZFS. ZFS is also "cleaner" in my opinion, in both design and UI. Oracle funding most btrfs development also raises a question of btrfs momentum now that they own ZFS and Solaris.
Re:Freedom ain't free (Score:3, Insightful)
It's not hard to be incompatible with the GPL. The GPL prohibits a lot of actions. Good for some, not for others.
Re:Freedom ain't free (Score:3, Insightful)
That's not the GPL's fault (Score:4, Insightful)
That's not the GPL's fault. It's the fault of the IP lawyers who are dicing permissions exceedingly fine. The GPL is designed to guarantee certain freedoms at the cost of others. It does its job very well, and is well architected with a lot of forethought considering we're only on version three after 21 years. At least one of those two revisions can be blamed not on the faults of the license but on the changing legal and IP environment.
Believe it or not once upon a time if you wrote some code somebody found interesting you just sent it to them. No patents. No copyrights. No approvals from management or legal. You just sent it, happy that someone else might benefit from not redoing the work you'd done once already. The idea of profiting from the derivatives they might make, or the derivatives of the derivatives, was simply not an idea that would occur to a normal person. If you had suggested such a thing at that time we'd have thought it hilarious.
And now I have to point to the onion on my belt, which was the fashion in my day.
Re:That's not the GPL's fault (Score:4, Insightful)
Yes, it is the GPL's fault. The CDDL is a per-file license. It places absolutely no restrictions on what other code can be combined with it in other files.
As the CDDL is deliberately GPL incompatible, had there not been any other issues, one can assume that Sun would have added 'may not be distributed together with GPL licensed code'. The CDDL/GPL incompatibility was on purpose, it was a feature asked for by Solaris engineers. Had the Linux kernel been BSD licensed, the CDDL would have been made incompatible with the BSD license.
Generally, fault implies some form of control over the issue. Under the circumstances, the only party with any control in this case would have been Sun, and as they would have redesigned the license until it was not compatible, it's quite obvious where any 'fault' should be assigned.
And unless the Oracle buyout has changed some attitudes within Sun for the better (heh), it's also quite naive of KQ Infotech to believe that Sun/Oracle would not go after them for violating the point of the license, as opposed to the actual text of the license (assuming any wider distribution). Standing is hardly a necessary prerequisite for a company of Oracles size to grind a small company into dust in the courts (and both Oracle and Sun would have standing as kernel contributors to sue any distributor of ZFS+Linux kernel combo).
Personally I can't say I consider it either a big loss or much to complain about. ZFS was a huge (HUGE) deal for Solaris, considering the painfully anemic storage stack it had in disksuite+ufs, but for any OS with a more modern volume management and file system stack it merely boiled down to a few nice features and some drawbacks, depending on your underlying storage architecture (SAN capabilities, etc).
Re:Freedom ain't free (Score:2, Insightful)
Ohwell, Oracle can always run their database on Linux once they managed to kill Solaris. And by then ZFS on Linux will be good for them ;)
Re:Freedom ain't free (Score:3, Insightful)
The GPL prevents redistribution of more free combinations as well.
...more free combinations which allow further redistribution of less free combinations, to be more exact.
Without this, GPL would be rather pointless. If somebody wants to keep their code free, and by extension, allow all future users of the code have certain freedoms that come with having the source code (what freedoms exactly, depends on GPL version, due to tivoization [wikipedia.org]), then that's what is needed.
Freedom (of any kind) is not black and white thing, nor is it one-dimensional scale.
Freedom to take freedom away sure is a freedom, but I can see why some would want to restrict that freedom when it's about something they've created and want to remain free.