ZFS sounds like a great filesystem and I'm glad it won't be just an option but used by default. If it was an option, people would never use it. I think disk IO is one of OS X's biggest problems with using laptop drives so hopefully this will go a little way to relieving it.
ZFS can be the boot volume. Some of the people I interact with in the Open Solaris community are already booting from ZFS on PCs by messing with the GRUB bootloader. My understanding is that the Mac's Extensible Firmware is much more flexible than GRUB, so it should be pretty easy to boot from ZFS on a Mac (in theory).
Xserves no longer ship with a RAID card option (!). This makes me think that ZFS is in the pipeline in some way, since ZFS (on a fast, 64-bit machine) is a more than adequate replacement for RAID-5, and has the advantage of not being tied to manufacturer-specific hardware.
ZFS is also endian-agnostic, which means that it can internally handle data no matter whether the system's processor is little-endian byte order (x86) or big-endian (PPC). That would be a pretty nice think for Mac users, I would think.
While it's true that you can add any device you like to a zpool, you CAN'T add a device to a RAID-Z pool and have it be part of the parity array (at least, you can't do this yet). So the extra disk you slap onto a pool that is configured to be RAID-Z won't get protected by parity - you'll have a sort of RAID-Z/JBOD pool at that point.
There IS and option when creating a stripe (which you CAN add to very flexibly) to specify that all bits be written between 1 and 3 separate times to the pool. I think ZFS will try to write each copy to a separate physical device, if possible.
Even if Schwartz is wrong (and I think he's an even more impressive CEO than Jobs), I'd at least recommend grabbing a copy of Solaris 10 or Open Solaris and installing it on an old PC or and Intel Mac and checking it out ZFS that way.
I would think that ZFS would allow you to have a virtual filesystem that crosses physical boundaries of hard drives.
I ran out of space on my main/boot harddrive because of video. With ZFS, I think I should be able to extend the filesystem so that it looks like all my videos are under my own username/home directory but actually are on 2 separate drives.
Any insights on how ZFS might benefit users?
As stated it works with virtual file systems. I would assume that HSF+ could be one of those.
It appears that ZFS supports forks/streams, so it should be possible to emulate most of the legacy features of HFS+.
The only thing I'm curious about is whether there's any way to implement the HFS+ feature of referencing a file by its file ID instead of its path, using FSRefs or aliases, so that your reference follows the file even if it is moved in the file system. Is this possible in ZFS, or is it something we'll have to say goodbye to (which would be a shame, because I don't know of any other OS that can do this)?
The simple creation of snapshots and clones of filesystems makes living with ZFS so much more enjoyable. A snapshot is a read-only point-in-time copy of a filesystem which takes practically no time to create and uses no additional space at the beginning. Any snapshot can be cloned to make a read-write filesystem and any snapshot of a filesystem can be restored to the original filesystem to return to the previous state. Snapshots can be written to other storage (disk, tape), transferred to another system, and converted back into a filesystem.
I know nothing about filesystems, but I know the chazm that always exists between my Power Mac and my PowerBook is incredibly annoying. I've repeatedly said that Apple should allow us to easily sync our portables (primarily Mail) with our main computer.
Is that part of what this point is referencing? Will ZFS make it easier to sync multiple machines?
Won't time machine either rely on ZFS to make it work, or at least, to make it work much better?
Time Machine is a relative back-up system. It backs up everything once, then it backs up changes on top of that. So if I back up daily, then the second back-up will only back-up things that I changed.
ZFS would let time machine work block-by-block instead of file by file. Imagine if I have my photo library backed up. If it's in a file-by-file set-up, then when I change one tiny little thing (tags, name, whatever), it will re-backup the whole photo (1-2 MB). Using ZFS, it'll re-backup a block (like a few KB or so).
With regards to the booting issue, Macs already use a separate boot partition of some sort, don't they?
THINKS "remember that I shouldnt believe everything I read (particularly on the internet)
Remember that Steve shuts off and shuts down pre announcing people, so even if it were true, he could kill it.
Remeber that its REALLY too close to WWDC for him to pull the plug on ZFS.
Conclusion = Revert to original first gut feeling on reading this.
OH MY GAAWD!
If true I am SO over the moon happy, there is VERY little I have read online that has made me want to join hands with everyone and dance around the room singing.
Maybe I'll get round to that, when it comes out of Steves mouth.
We know that Apple has jammed ZFS + 'dtrace' + Java from
Sun into MacOS X for a while now. Perhaps a "secret feature" from
the near/medium-term future is to just replace the whole damn
kernel with (soon to be GPL3) Solaris. Natch, this will only
run on Intel-based non-legacy Macs.
Basically every OS runs Java. When something doesn't have a Java runtime, that's notable.
There's no way that you can effortlessly swap out a kernel. There's a lot of interfacing between the kernel and the userspace.
No offense, but by your logic, Apple uses bash, gcc, and about 80 other GNU utilities. Maybe they're switching to Hrud. Or they use OpenSSH, maybe they're switching to a pure OpenBSD kernel. Sorry, but that doesn't really hold weight.
Not to mention, what does the OpenSolaris kernel have that Darwin doesn't?
I meant that it's "relative" in the sense of it makes incremental backups in relation to the first back-up. So if you have a 50 GB filesystem (say, 10,000 files) backed up on Monday, then if you change 50 files, then it only backs up those 50 files, resulting in a total back-up size of like 50.1 GB (actually, there's compression, so it'd be smaller). In an absolute backup system, you'd have 50 GB + 50 GB = 100 GB of data over those two days.
Comments
I like HFS+ but it's carrying a *lot* of legacy code (6800 code emulated into powerPC emulated to x86).
That sentence makes absolutely no sense.
"I'm gonna kill you, Schwartz!" -Steve Jobs
NO SJ doesn't do that. He'll just say, you're not being professional and creative enough, let's adopt a black turleneck attire shall we....
ZFS sounds like a great filesystem and I'm glad it won't be just an option but used by default. If it was an option, people would never use it. I think disk IO is one of OS X's biggest problems with using laptop drives so hopefully this will go a little way to relieving it.
This is great news, Filesystems like ZFS are essential for a future where storage is online, all-the-time.
Hmmmm. . . . Do I smell a .mac upgrade in the works? Steve did mention it last week. . . .
~Matt
Xserves no longer ship with a RAID card option (!). This makes me think that ZFS is in the pipeline in some way, since ZFS (on a fast, 64-bit machine) is a more than adequate replacement for RAID-5, and has the advantage of not being tied to manufacturer-specific hardware.
ZFS is also endian-agnostic, which means that it can internally handle data no matter whether the system's processor is little-endian byte order (x86) or big-endian (PPC). That would be a pretty nice think for Mac users, I would think.
While it's true that you can add any device you like to a zpool, you CAN'T add a device to a RAID-Z pool and have it be part of the parity array (at least, you can't do this yet). So the extra disk you slap onto a pool that is configured to be RAID-Z won't get protected by parity - you'll have a sort of RAID-Z/JBOD pool at that point.
There IS and option when creating a stripe (which you CAN add to very flexibly) to specify that all bits be written between 1 and 3 separate times to the pool. I think ZFS will try to write each copy to a separate physical device, if possible.
Even if Schwartz is wrong (and I think he's an even more impressive CEO than Jobs), I'd at least recommend grabbing a copy of Solaris 10 or Open Solaris and installing it on an old PC or and Intel Mac and checking it out ZFS that way.
The Steve is powerful. Next, when you go to the Sun site you'll find a pottery enthusiast page.
Do not pre-announce the Steve's magics.
This was out there in December. So Jonathan isn't really leaking anything.
I would think that ZFS would allow you to have a virtual filesystem that crosses physical boundaries of hard drives.
I ran out of space on my main/boot harddrive because of video. With ZFS, I think I should be able to extend the filesystem so that it looks like all my videos are under my own username/home directory but actually are on 2 separate drives.
Any insights on how ZFS might benefit users?
As stated it works with virtual file systems. I would assume that HSF+ could be one of those.
Won't time machine either rely on ZFS to make it work, or at least, to make it work much better?
The only thing I'm curious about is whether there's any way to implement the HFS+ feature of referencing a file by its file ID instead of its path, using FSRefs or aliases, so that your reference follows the file even if it is moved in the file system. Is this possible in ZFS, or is it something we'll have to say goodbye to (which would be a shame, because I don't know of any other OS that can do this)?
Stupid question:
Won't time machine either rely on ZFS to make it work, or at least, to make it work much better?
Not stupid at all. I'm beginning to think that Apple knew they'd be moving to ZFS.
1. Time Machine has "snapshots" written all over it.
2. The Xserve had no RAID card which was WTF?
3. No XSAN update in a while (longshot but perhaps it's going to be delivered with ZFS supports)
10. Clones with no ethical issues
The simple creation of snapshots and clones of filesystems makes living with ZFS so much more enjoyable. A snapshot is a read-only point-in-time copy of a filesystem which takes practically no time to create and uses no additional space at the beginning. Any snapshot can be cloned to make a read-write filesystem and any snapshot of a filesystem can be restored to the original filesystem to return to the previous state. Snapshots can be written to other storage (disk, tape), transferred to another system, and converted back into a filesystem.
I know nothing about filesystems, but I know the chazm that always exists between my Power Mac and my PowerBook is incredibly annoying. I've repeatedly said that Apple should allow us to easily sync our portables (primarily Mail) with our main computer.
Is that part of what this point is referencing? Will ZFS make it easier to sync multiple machines?
Stupid question:
Won't time machine either rely on ZFS to make it work, or at least, to make it work much better?
Time Machine is a relative back-up system. It backs up everything once, then it backs up changes on top of that. So if I back up daily, then the second back-up will only back-up things that I changed.
ZFS would let time machine work block-by-block instead of file by file. Imagine if I have my photo library backed up. If it's in a file-by-file set-up, then when I change one tiny little thing (tags, name, whatever), it will re-backup the whole photo (1-2 MB). Using ZFS, it'll re-backup a block (like a few KB or so).
With regards to the booting issue, Macs already use a separate boot partition of some sort, don't they?
Sun into MacOS X for a while now. Perhaps a "secret feature" from
the near/medium-term future is to just replace the whole damn
kernel with (soon to be GPL3) Solaris. Natch, this will only
run on Intel-based non-legacy Macs.
THINKS "remember that I shouldnt believe everything I read (particularly on the internet)
Remember that Steve shuts off and shuts down pre announcing people, so even if it were true, he could kill it.
Remeber that its REALLY too close to WWDC for him to pull the plug on ZFS.
Conclusion = Revert to original first gut feeling on reading this.
OH MY GAAWD!
If true I am SO over the moon happy, there is VERY little I have read online that has made me want to join hands with everyone and dance around the room singing.
Maybe I'll get round to that, when it comes out of Steves mouth.
Time Machine is a relative back-up system.
You think?
We know that Apple has jammed ZFS + 'dtrace' + Java from
Sun into MacOS X for a while now. Perhaps a "secret feature" from
the near/medium-term future is to just replace the whole damn
kernel with (soon to be GPL3) Solaris. Natch, this will only
run on Intel-based non-legacy Macs.
Basically every OS runs Java. When something doesn't have a Java runtime, that's notable.
There's no way that you can effortlessly swap out a kernel. There's a lot of interfacing between the kernel and the userspace.
No offense, but by your logic, Apple uses bash, gcc, and about 80 other GNU utilities. Maybe they're switching to Hrud. Or they use OpenSSH, maybe they're switching to a pure OpenBSD kernel. Sorry, but that doesn't really hold weight.
Not to mention, what does the OpenSolaris kernel have that Darwin doesn't?
You think?
I meant that it's "relative" in the sense of it makes incremental backups in relation to the first back-up. So if you have a 50 GB filesystem (say, 10,000 files) backed up on Monday, then if you change 50 files, then it only backs up those 50 files, resulting in a total back-up size of like 50.1 GB (actually, there's compression, so it'd be smaller). In an absolute backup system, you'd have 50 GB + 50 GB = 100 GB of data over those two days.