Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I seriously doubt Apple can make something has "all the advantages of ZFS plus a few more goodies." in only a couple years. They have been trying to port ZFS for a couple years already and still were not done. Starting from scratch is a 10X harder problem.

I've followed the development of several new file systems. Two years is "way not enough" time. Many of the projects fail. You can't even be sure you will finish. I think Microsoft worked for many years on some kind of DBMS based file system to replace NTFS and it seems they just gave up

The big catch isn't that it takes forever to develop the filesystem code per se... but it does take forever to refine it.

App crashed? Lost maybe an hour of work.
OS crashed and wrote bad output through the FS driver? Lost a drive, maybe, but your backups should be fine.
FS driver behaving badly? You have the potential to lose it all, even your backups you made off that machine.

The level of danger with shipping an FS out the door with bugs is much, much bigger than even an OS kernel.

So projects like this take a ton of time because they need a ton of time to make sure it won't fail in hilarious ways (at least, hilarious to those who didn't lose years of data in the blink of an eye).
 
The main and really good features (volume management, filesystems, snapshot management) are the very useful for enterprises. If you have any things that are only useful for enterprises, please enlighten me.

No ..my point being Apple is not an Enterprise vendor so snapshots really aren't needed at their level (consumer, SMB). Apple needs a way to preserve more metadata within the filesystem. Right now the system is kludgey with many disparate places to store or retrieve data.
 
OS X was secretly running on intel chips YEARS before they decided to switch over. They were working on it in the background, 'just in case.'

What makes you think they're just starting on their next file system now? Just because they were playing around with ZFS? Apple has never been a company to put all their eggs in one basket.

They likey are working on something else. They'd be dumb not to be but I doubt they started many years ago. These things take "forever" to create. If we do see a new file system in 10.7 it will not be comparable to ZFS. Your example of the Drobo is more like what is needed anyways or maybe like the file system used by Netapps. The netapp system has a time machine-like snapshot abilty.

Getting Mac OS X running on Intel was easy. The kernel and much of the other software was designed for portability. When Next snagged BSD they had to port it the the MC68000 Next Cube and then later when Apple snagged Next OS they had to port it to PPC and all the while Next and Apple were exchanging code with the BSD folks who were mostly using Intel machines. Years before Jobs and Woz built the Apple II the Unix code had a tradition of being ported to all kinds of hardware, The guys at Next and Apple simple kept that tradition after they adopted the code base.
 
May be they will use ext4.

God I hope not. Ext4 is version 4 of a filesystem from 1992! It may have some added features since Ext1, but it's hardly a spring chicken anymore.

BtrFS or a brand new FS ftw! Something with logical volume support, integrated parity and some frakking undelete please!!!
 
I wish someone could explain in simple terms why this ZFS is the biggest thing that everyone should jump on since the invention of the transistor.

How does ZFS improve things for the average Mac user over the current system?
 
Predictable

Well, Oracle shafted Apple by not delivering a release on x86 OSX. Apple screws Sun/Oracle by not adopting what is arguably a very nice filesystem technology. Too bad because we all loose in the end. Oracle 11g would have been an awesome combination for Xserve/XSAN/XFS.

To be fair though, Oracle is being irrational about Sun and their renews "commitment" to Sparc processor development simply drives home the fact that they don't "get it" and are probably about to enter a period of decline and irrelevance.

@hugodrax: to answer your question. XFS offers allot of midrange storage functionality that used to come with a big helping of complexity. Much like not having a viable relational database system like Oracle, Apple can't really make much progress into the midrange server market where OSX Server could potentially thrive.
 
I can't say I'm surprised. Maybe I'm way off base here, but it seems like there are only a few features here that would benefit the consumer. Storage pools and copies of previous file versions. Both of these features are present in other, more consumer-friendly technologies: The Drobo and Time Machine.

Given this, ZFS seems like it's designed as a storage file system, not necessarily an OS one. I'm not one to follow File Systems that closely, but it seems like HFS+ is a really good one. If re-written to take advantage of some more modern technologies (like SSD drives, storage pools and previous versions) it could be great.
 
Apple hired Dominic Giampaolo years ago folks. Dom wrote the FS for Be OS.

Apple was mainly kicking the tires with ZFS but they in now way ever told anyone that they planned on making ZFS their default FS.

I've been waiting since I gave up BeOS for dead for that Giampaolo hire to turn out something cool—hopefully we'll see it sooner rather than later. BeFS was one of the coolest things about BeOS, and that was more than ten years ago.

Show us your stuff, Dom! :D
 
Apple creating a brand new filesystem would be terrible

For one, creating a file system is hard. Really hard. Especially if you have such a large install base like HFS+ and you need to migrate data over. If it was easy, apple would have probably been using ZFS much earlier and just licensed it from Sun.

Also, if they started creating a filesystem from scratch right now, it would take years before it would be stable enough for everyday users to use it. Look at Btrfs (http://en.wikipedia.org/wiki/Btrfs) for linux for proof of that. Btrfs has many of the same goals that ZFS had but without the licensing issues of CDDL. They are still testing Btrfs in the linux kernel and it probably won't be considered stable until 2010 or 2011. Creating a brand new FS now would mean that it wouldn't see the light of day for non-server macs until ~2015.

So, what's my vote for a new next gen apple filesystem? Hammer. (http://en.wikipedia.org/wiki/HAMMER) It has many of the same goals that ZFS had but it's got a BSD style license.
 
Sun has set

I loved the idea of what ZFS promised while realizing that Sun was in no way capable of delivering on that promise, their singular lack of market understanding was the problem and wont be improved by Larry absorbing them in to Oracle, Apple was the only bright spot in the idea as far as I was concerned with a chance of making it go mainstream.

What now for a new file storage system?
 
And yes, I'm mainly talking out my ass but doesn't it make sense??

Whether it makes sense or not, when on forums I generally assume everyone is "talking out of their *ss". I don't think there's any need to explicitly admit it. :p

I'm not too surprised by this announcement. Even before the Oracle buyout, Sun often seemed to be wrestling with itself - especially when it came to their relationship with open source (not that Apple is always faultless in this regard). Sun would say "xxx is going to be GPLed/BSDed/license-du-jour'ed", and then it'd get pushed back, disavowed, re-announced, etc. time after time after time.
 
I'd rather have Apple custom developer their own file system.

If they couldn't come up with their own operating system, cough Copeland, why do you think they can do a file system ? Apple's ZFS file system has taken as long to gestate as Longhorn did for MS. MS has been working on a new file system for 15 years and they still can't get it right.
 
If you think HAMMER will be stable before BtrFS, I have this be-you-tea-ful bridge to sell you, in a prime location in the middle of London!

It's not only time that is important, also number of programmers, testers and how much of it's resources Apple is willing to make vailable for the project. 2 or 3 years is not an impossible deadline ;)
 
May be they will use ext4.

I had the same thought.

If ZFS is a no go, why not new generations of ext?

Apple might be able to improve HFS, or write a new one... but a somewhat more universal, less proprietary, but still GOOD file system seems like a better option than continuing this closed-file-system p!$$!ng match.

Network storage, distributed systems, and users who use multiple machines, sometimes across platforms... it makes more and more sense for next-gen performance to be readable and writeable by other operating systems, even if the OS itself 'requires' or prefers a specific file system to be resident on...

If nobody can seem to get new file systems right... maybe they should pool resources, and develop an industry standard file system framework, perhaps based on advances in neuro-science or something... Use biology as an inspiration as well as mathematics, to design a way to sort, move, and manage recorded digital data at a high rate of speed with high accuracy, and robustness.

Maybe there is a case to be made for a standard set for that, rather than multiple pools of R&D trying to struggle away at it, and not making much progress. Maybe collaboration and less emphasis on protecting that particular proprietary technology would be good on this one topic.

The OS can be specific or proprietary on top of that, if companies wish. There is still plenty of "this is mine, that is yours" to go around otherwise.
 
It's not only time that is important, also number of programmers, testers and how much of it's resources Apple is willing to make vailable for the project. 2 or 3 years is not an impossible deadline ;)

Sure, unlike Oracle. Do you realize BtrFS is already installable with apt-get in Ubuntu?

BTRFS is GPL and that makes it a no go.

I am making popcorn in anticipation of this explanation!
 
I'm shocked by this. I agree that it would almost certainly be the Oracle/Sun tie-up. What a shame!

There is no reason why ZFS isnt the perfect filesystem/storage management solution for OS X.

All macs now have (powerful) Intel processors. Apple could have hidden the complexities of ZFS from the user. Apple could have optimised and reduced the overhead of ZFS using Snow Leopard technologies (specifically, Grand Central Dispatch and OpenCL). To make maximum use of multicores and graphics processing.

I cannot see Apple being able to come up with anything as good as ZFS. Its crazy since ZFS is already stable.
 
I am making popcorn in anticipation of this explanation!

Standard problem with the GPL and commercial licensed software, "if you link - you must release your code under GPL". Apple might not want to do that. A File System is a much different beast then including some GPL programs.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.