More on Apple and ZFS, speculation on iOS

More on Apple and ZFS, speculation on iOS

Following up on the information iMore received yesterday on potential ZFS hooks in iOS 5.1, a reliable source let us know that Apple has been investing heavily in ZFS appliances from Oracle. (Whether or not this will be made public is unknown; it doesn't seem to be right now.)

Also, several of our readers familiar with ZFS wrote in to speculate as to why Apple may be exploring ZFS for use on iPhone and iPad. Broadly, it may have something to do with ZFS' snapshot technology. ZFS snapshots can be easily compressed and encrypted, de-duplicated, and can transfer delta (bit differential) updates between cloud and device. The snapshots can also work like Time Machine, easily and efficiently providing incremental backup and restore capability from an hour ago, a day ago, a week ago, or more, asynchronously and block by block. Here's what Wikipedia says about ZFS snapshots:

An advantage of copy-on-write is that when ZFS writes new data, the blocks containing the old data can be retained, allowing a snapshot version of the file system to be maintained. ZFS snapshots are created very quickly, since all the data composing the snapshot is already stored; they are also space efficient, since any unchanged data is shared among the file system and its snapshots.

Writeable snapshots ("clones") can also be created, resulting in two independent file systems that share a set of blocks. As changes are made to any of the clone file systems, new data blocks are created to reflect those changes, but any unchanged blocks continue to be shared, no matter how many clones exist. This is an implementation of the Copy-on-write principle.

Snapshots is only one of the features of ZFS (Zettabyte Files System), the advanced file system created by Sun, now owned by Oracle. Apple included a partial port of ZFS in earlier versions of Mac OS X but pulled it before the official release. It's a file system particularly well suited for cloud environments and storage pools, which might be why Apple is looking at it again in the post iCloud.

Apple is an incredibly private company that works hard behind the scenes in order to present a beautifully packaged, highly polished product to consumers. No doubt there is a lot of really cool, really geeky technology at the core of iCloud, for example, but all most of us see are elegant interface elements in Settings.app that, when tapped, all but transparently backup and restore our iPhones or iPads.

If indeed Apple is planning on using ZFS to make iCloud backups and restores better, it will probably be invisible to end users. One day, iCloud will simply work better, faster, and even more reliably.

More: Wikipedia

Have something to say about this story? Share your comments below! Need help with something else? Submit your question!

Rene Ritchie

Editor-in-Chief of iMore, co-host of Iterate, Debug, Review, Vector, and MacBreak Weekly podcasts. Cook, grappler, photon wrangler. Follow him on Twitter and Google+.

More Posts

 

-
loading...
-
loading...
-
loading...
-
loading...

← Previously

iMore editors app picks of the week for February 4, 2012 [Giveaway]

Next up →

How to follow the Super Bowl on your iPhone and iPad

Reader comments

More on Apple and ZFS, speculation on iOS

15 Comments

I don't understand: what's wrong with iCould at present? You imply it could be better and faster--in what way is its speed a problem right now? Also, you imply it's unreliable--in what way? I'm not trying to be a smart aleck or anything: I'm new to Apple products and i haven't noticed any problems with iCloud (that i can remember, anyway), so I'd like to be brought up to speed (and I'd like to be more reliable, too.....). Thanks!

Hi Martin,
Thanks for commenting! We're not implying anything about iCloud not being fast, good, or reliable, but the truth about computing technology is that everything can get better. iPhone 4 was fast, so saying iPhone 4S is faster doesn't suddenly make iPhone 4 slow.
When it comes to online especially, there's so much innovation going on that things will always be getting better. Just like Apple will make an iOS 6 and an iPhone 5 eventually, they'll likely keep adding features to iCloud. And that's what we all want, I think.

I think you will find that this snapshot technology was actually invented by a company called StorageTek (Louisville, Colo.) and used in their Iceberg Disk Array System.
Sun Microsystems bought (and pretty much destroyed) StorageTek many years ago ... at least some of STK's legacy lives on.

Yes, it's inevitable Apple will release a new iPhone eventually. But it's not inevitable it will be called "iPhone 5". The reasons why iMore and the rest of the blogosphere are calling it that, I don't understand at all. What am I missing?

For God sake man! Does it really matter what the next iPhone, iPad or whatever iDevice comes out will be called? Seriously who cares! All that matters is the quality of the design, hardware and user experience of the device.

I'm not sure why using the term "iPhone 5" upsets some people. It's just a name. Internally at Apple, the next iPhone is "iPhone 5,1" (iPhone 3G wasn't the second generation, it was iPhone 1,2 or a variant of the original iPhone).
Apple could call the next iPhone anything they want. iPhone X. iPhone RED. iPhone Banana. Right now, Apple has had two iPhone models with 3 in it -- iPhone 3G and iPhone 3GS, and two models with 4 in it -- iPhone 4 and iPhone 4S. There's every chance the next two will be iPhone 5 and iPhone 5S.
Either way, it's nothing to get upset about.

The rumours previously said Apple were busy developing in-house their own new FS for Mac OS X after the collapse of their licensing negotiations for ZFS with Sun/Oracle. If they're going back to ZFS for iOS then does that mean Mac OS X will also use ZFS and the internal new FS project is dead?
I can't see Apple using ZFS in the iCloud backend and iOS clients while building a completely different FS for Mac OS X. Macs use iCloud too and if iOS is going to use ZFS then surely Mac OS X must be going to also.

The main design goal of ZFS was, and still is, to maintain data integrity. Its file system code can recover from many types of data degradation, such as "bit-rot," alpha particle hits, ghost-writes, etc.
So, why would data integrity be more of an issue now than it was before? Because more and more devices use flash memory, which is subject to higher rates of data integrity deterioration than other types of storage. ZFS should be able to recover from memory cell degradation and failures.
Combined with Apple's purchase of Anobit, the Israeli flash memory company, Apple should be well prepared for even more extensive use of flash memory. Anobit's technology is apparently very good at spreading out read/erase wear across flash memory blocks. This reduces the average number of cycles per cell, which statistically should increase the usable life of flash memory. Combining that with ZFS' superior error recovery would give Apple a very robust flash memory-based file system for future devices.

ZFS rocks but we're not likely to see it on mobile devices. ZFS is an excellent volume manager and file system with great features, indeed it can detect data corruption due to checksumming every block of data. To recover from data corruption you will need another copy or some way to reconstruct it, I.e mirror or raid with parity. If ZFS detects corruption it may just stop dead, potentially saving further corruption till someone gets eyes on it, mdb, zdb etc.
I don't think we will have iPhones with 32g+32g storage soon. I suppose they could slice the storage in half and create a zpool mirror on the two halves ;-)

ZFS can actually store 2 (or even three) copies of a file on a single volume using the copies parameter. This can be done on a selective volume by volume basis. ZFS will recover transparently should one copy be bad, and fail should both copies be bad.
This will save you from bad blocks, but not from a bad flash drive/device.

Hello from the http://MacZFS.org/ community (the free ZFS on Mac OS, which has always existed for years). I'd sure like to get some substantiation to this. Even this screenshot is a rumor. It's just text. And the presence of zpool binaries doesn't mean that they function. Do we know where it came from? Was it an internal-only build, or was it ever released to the public? Has it appeared in any other releases? Is it able to create and maintain zpools?

I maintain the MacZFS FAQ, so I made an entry for it just to address the rumor. Cool stuff!

http://code.google.com/p/maczfs/wiki/FAQ#Does_ZFS_run_on_iOS?

Remember to paste the whole link there, in case this web site doesn't catch the question mark.