Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
ZFS To Become Default File System In Leopard
- Thread starter MacRumors
- Start date
- Sort by reaction score
ZFS sounds like a step in the right direction. Roll on Leopard
I think ZFS is the final step in File Systems when it comes to limitations ^^
naw, something better will always come along, but ZFS is the top of the line right now, imo.
Can Windows XP or Vista read ZFS?
If so, how will this affect the "safety net" surrounding the MacOS Partition for Boot Camp or Parrallels users?
I thought that the MacOS partitions were safe from Windows viruses because the Windows OS's couldn't natively read the drives formatted for Mac. Will this still be the case?
If so, how will this affect the "safety net" surrounding the MacOS Partition for Boot Camp or Parrallels users?
I thought that the MacOS partitions were safe from Windows viruses because the Windows OS's couldn't natively read the drives formatted for Mac. Will this still be the case?
I don't need to scale my storage down. And your almost right. You can swap drives if you want to grow with new drives:
zpool replace [-f] pool old_device [new_device]
Replaces old_device with new_device. This is equivalent to attaching new_device, waiting for it to resilver, and then detaching old_device.
The size of new_device must be greater than or equal to the minimum size of all the devices in a mirror or raidz configuration.
If new_device is not specified, it defaults to old_device. This form of replacement is useful after an existing disk has failed and has been physically replaced. In this case, the new disk may have the same /dev/dsk path as the old device, even though it is actually a different disk. ZFS recognizes this.
Replaces old_device with new_device. This is equivalent to attaching new_device, waiting for it to resilver, and then detaching old_device.
The size of new_device must be greater than or equal to the minimum size of all the devices in a mirror or raidz configuration.
If new_device is not specified, it defaults to old_device. This form of replacement is useful after an existing disk has failed and has been physically replaced. In this case, the new disk may have the same /dev/dsk path as the old device, even though it is actually a different disk. ZFS recognizes this.
Also, if you use ZFS in a redundant state you can purposely fault a drive and replace it that way. The only thing you can't do is seamlessly scale your capacity down. But they're working on it.
I don't get what's so new about this. Hasn't JBOD been around since the 80's? I've been doing a similar thing (expanding a windows "spanned" volume) since 2005, but the feature is present back in windows 2000 and pretty simple to use...
Eh, I don't think there's any such thing just yet. The idea of needing a terabyte at home seemed crazy at one point. The importance of metadata has skyrocketed over the past ten years, and it wasn't something that everyone was predicting would happen back in the 80s. It's hard to say what the needs of future OSes will be. For OS X, for instance, Spotlight and Time Machine are the two technologies that most drive changes in the way the FS should work. Spotlight came out just in Tiger and TM isn't even out yet!
But it does seem like a really good option right now, with a good amount of room for growth.
BTW, Elwe, thank you for your very informative post directed to me, a number of posts above!
They already have this. It's called RAID-0.
The thing about ZFS is: you can add more drives to the array later without formatting ... unless I'm totally on crack.
Not necessarily. OSX could easily partition a small section of the boot disk into HFS+ during installation and use that for booting. Then the rest of the disk would be in ZFS. The boot section could even be a hidden partition.
I like the ability to add new disks into a seamless ZFS pool. The continuous error detection sounds nice as well. If TimeMachine can take advantage of all I read about the incremental write abilities to keep the snapshot and backup sizes small, that would be great. I think that would be the killer feature of ZFS for normal users.
I think the previous poster meant that it is impressive that Apple is almost ahead of Sun in implementing Sun's FS!
JBOD has no redundancy and requires extra hardware to implement well. Also, it doesn't give you the performance of RAID 0. The only thing JBOD gives you is easy expansion of data storage, but it doesn't provide any kind of protection or performance enhancement and it requires additional hardware. I've not seen a software JBOD offering but they may exist.
ZFS takes a lot of good, old, ideas and puts them together with some novel concepts. Its the overall capabilities of ZFS that are cool, not any one feature isolation.
Only one person has mentioned this so far, but the big news here is that they must've managed to get around the limitation that ZFS hasn't been able to be booted off until now (or very recently).
That's very interesting.
That's very interesting.
Well, yeah.
Still nifty. Built-in, seamless RAID capability with almost no work required.
In some ways, except that Drobo isn't as broadly useful as ZFS. Perhaps if Drobo made the software that makes it work available in some general way other than as an appliance they could be accurately compared.
I thought he was alluding to the disclaimer that ZFS cannot be used as a boot drive (at least for Solaris), but OSX would have this ability. I think Leopard does not have to be able to boot from ZFS, but still advertise ZFS as the default.
As far as I am aware, Intel macs don't support Classic, but PPC ones do. I haven't heard anything suggesting that Classic support is being dropped from Leopard on PPC macs... Did I miss something?
The win2k implementation doesn't require any additional hardware. If windows can see the drive, then you can add it to the stripeset. The performance gain would be nice I suppose, but in that case, this is more like RAID 0 + JBOD. That's pretty nice, but nothing earthshattering. And from what people in this thread have said, if a drive dies, then a drive dies, and ZFS isn't going to bring back your data.
I'm excited about this, but honestly, it's not that big of a deal.
No. Microsoft originally intended to do their own next-generation filesystem (the whole sorry WinFS affair) this time around, but it wasn't cooked enough.
There are quite a few independently written IFS (filesystem drivers) for windows, so there probably will be support for it sooner or later (but you'd have to install that yourself).
No real change. It is possible for Windows to access Mac partitions if something like Macdrive is installed (but again, this isn't something included by Microsoft).
A determined program can muck with a partition even if there isn't built-in OS support for its filesystem. It's best to keep the defenses up on the Windows side, and not rely too much on that foreign filesystem thing as protection.
Sound great to me. I think it will be a necessary edition of Leopard, at least to have Time Machine fully functional (based on the error correction and compression abilities of ZFS)
And I am CERTAIN apple will make the transition painless for existing users, with no additional hardware. Heck, we got through the intel transition relatively flawlessly, and that seems a lot more intensive than this.
Leopard is turing out to be quite a strong cat!
And I am CERTAIN apple will make the transition painless for existing users, with no additional hardware. Heck, we got through the intel transition relatively flawlessly, and that seems a lot more intensive than this.
Leopard is turing out to be quite a strong cat!
Sounds like a good thing.
sounds like a very good thing! this is great news....very glad to hear this. looking forward to Leopard
ZFS provides data redundancy either via mirrored pools or RAID-Z. Mirrored pools provide for checksum verification between the pools and mean when a drive fails the data can be easily recovered. RAID-Z is like RAID 5 without the performance hit on write since ZFS uses copy-on-write. ZFS isn't earthshattering, but it's a lot more than just an iterative development in terms of filesystems. Sun explicitly states that running ZFS in a non-redundant configuration is a bad idea, although possible. You have basically three ways you can run ZFS:
1.) Single Pool - RAID 0 performance with some very basic data integrity via checksums. Much easier to expand than RAID 0 since ZFS dynamically restripes the data. Also gives you the ZFS snapshot feature. - This could be nice for home users who still plan to keep some other kind of backup.
2.) Mirrored Pools - RAID 10 type functionality but with the ability to scale easier and much better data redundancy via mirroring and checksumming. Protects against checksums getting goofed up. - I like this solution for servers and workstations where data is critical.
3.) RAID-Z - RAID 5 type functionality without the performance hit. - Best for home users who want data integrity but want to use as much of their drives for actual data storage rather than redundancy.
Only the single pool has no redundancy although the checksums do protect against some hardware defects (bad sectors). Go checkout the ZFS screencasts and docs at OpenSolaris.org.