It would certainly make XServes more attractive. But they'd need a lot more than ZFS to really get them moving well. Apple is going to need to put some marketing behind them.
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
It would certainly make XServes more attractive. But they'd need a lot more than ZFS to really get them moving well. Apple is going to need to put some marketing behind them.
I sense a disturbance in the farce. Teh Steve is not pleased.
Yah know come to think of it he already has the back turtleneck. All he needs is a cloak. The RDF can already produce lighting bolts. We've never seen it since no one has pissed teh Steve off to that extent.
"Young fool. Only at the end do you understand!" *lighting* Bill G screams.
"Father please! Help me!" Woz turns to Steve. Turns back to Bill. Then throws Steve down the shaft of the newly constructed Death Apple.
PS- Somewhere out there John Siracusa just wet himself.
Yah know come to think of it he already has the back turtleneck. All he needs is a cloak. The RDF can already produce lighting bolts. We've never seen it since no one has pissed teh Steve off to that extent.
"Young fool. Only at the end do you understand!" *lighting* Bill G screams.
"Father please! Help me!" Woz turns to Steve. Turns back to Bill. Then throws Steve down the shaft of the newly constructed Death Apple.
PS- Somewhere out there John Siracusa just wet himself.
Is this a joint effort.
I haven't heard it mentioned so I'll just ask here. Is this a joint effort between Apple and SUN to marry ZFS to OS X, or is Apple going it alone with the ZFS open source?
If it's a collaboration bewteen Apple and SUN will the source code have to be published (Especially if it boots ZFS)?
I haven't heard it mentioned so I'll just ask here. Is this a joint effort between Apple and SUN to marry ZFS to OS X, or is Apple going it alone with the ZFS open source?
If it's a collaboration bewteen Apple and SUN will the source code have to be published (Especially if it boots ZFS)?
I don't see why anyone would want to do that.
Convenience. If you have 3 500GB drives, it can be a bit of a pain to find things since they can be on either of the 3 drives. If they're combined into one logical volume, it makes things easier since it's just one big drive and you don't have to worry about where everything's stored.
Of course, with 3 drives, it's still relatively easy to find things. But imagine an enterprise-grade file server with 100 drives, each 500GB. It would be a hell of a lot easier to do things if those 100 drives made up 1 logical drive. This is already possible with RAID. ZFS just has support for it built-in, with a ton of other improvements.
Right, I was only talking about the filesystem level. It's taken awhile for the unix stuff to catch up with Mac-specific quirks, and they're still not all there. Only in Tiger did various file utilities get the ability to deal with resource forks.
Since Tiger came out, I've been using 'mdfind' instead of locate. It seems to be case-insensitive, and has the added benefit of indexing the contents of the files (though that can be a drawback if it produces a huge amount of results -- but they can always be filtered down). I think I've used the Spotlight menu only a handful of times in the last two years, but I use mdfind quite regularly.
OS X Server did get case-sensitive HFS+ first (Panther Server), so there is a precedent for that sort of thing. Though I believe both got journaling at the same time (around 10.2.3, right?). It's all speculation until next week!
Actually, you might be able to. ZFS allows you to add devices which are actually normal files to a pool, it might work for files located on an NFS volume. I haven't tried this, and even if it works it certainly isn't recommended.
This if for Solaris of course. We don't know what changes Apple has made until 10.5 is released.
Excellent! I guess that answers my question about case-insensitivity. I wonder if this is the last hurdle in regards to supporting legacy (or "lazy" in the case of CS3) apps that assume certain HFS+ features? It never made sense to me why certain apps still failed to run on UFS filesystems when Apple added the ._filename convention for resource forks. Was it a flaw in this workaround, or was it something like case-sensitivity? Will ZFS now have all of the required workarounds or translations to convince any app that it's running on HFS+ or equivalent?
Should be interesting to find out.
That's true, but what about the firmware (Open Firmware for PPC, EFI for Intel)? I imagine it could be modified to support ZFS, but I certainly don't know that for a fact. The firmware needs to be able to read the filesystem in at least a rudimentary capacity so it can find the right file to boot from. Once the OS is in memory, it can then load the full filesystem drivers and go from there.
Note that the OpenSolaris post linked previous about x86 booting was presumably talking about booting Solaris on x86, not anything to do with Mac OS X. Solaris uses the open-source GRUB to boot on x86, whereas I believe Sparc-based Suns use Open Firmware (hmm, seems like we've seen that name before!) to boot. Most likely it was easier to tackle modifying GRUB first before doing it for Open Firmware, but I don't know!
I dunno... maybe in a server type situation. I do video work...this would seem like a nightmare for me. I have separate drives for separate things. Combining drives would be one of the last things I'd want.
I don't see why anyone would want to do that.
Why not? It makes things a lot simpler. Now you could have one smaller drive and one bigger drive. What if the smaller drive becomes full? Well, you have a problem then. And it doesn't really help if the other drive is 90% empty. With pooled storage that simply does not happen.
Consider this: You are doing photo-editing and you have your master-library on some HD. That HD then becomes full, and you decide to add another HD. What do you do to your library then? Split it to two different locations? Move the library to the new bigger drive, while leaving the existing HD unused? With pooled storage you simply add the new HD to the storage-pool and continue like nothing had happened. It's all totally transparent to the apps.
I really don't see how. Why would one storage-pool be a nightmare? You could still have separate folders to set things apart from each other. storage-pool eliminates the problem where some of your hard-drives are almost empty, while others are filled to the brim. It really is the way of the future, and just about only reason some people are against it, is legacy-thinking. we are used to separate drives. But that really doesn't have to change. Instead of separate drives, you could have separate folders.
Marc Hamilton No Longer Unequivocal
From http://daringfireball.net/:
Marc Hamilton No Longer Unequivocal
Marc Hamilton, VP for Solaris Marketing, in a comment following up on his
weblog entry about ZFS being announced as the new default for Mac OS X:
I don’t know Apple’s product plans for Leopard so it certainly wouldn’t be
appropriate for me to confirm anything. […] There certainly have been plenty
of published reports from various sources that ZFS is in Leopard, I guess we
will all have to wait until it is released to see if ZFS made it as the default, or
if they simply announce that it will become the default in a future release.
http://blogs.sun.com/marchamilton/entry/sun_s_new_modular_blade#comment-1181185661000
From http://daringfireball.net/:
Marc Hamilton No Longer Unequivocal
Marc Hamilton, VP for Solaris Marketing, in a comment following up on his
weblog entry about ZFS being announced as the new default for Mac OS X:
I don’t know Apple’s product plans for Leopard so it certainly wouldn’t be
appropriate for me to confirm anything. […] There certainly have been plenty
of published reports from various sources that ZFS is in Leopard, I guess we
will all have to wait until it is released to see if ZFS made it as the default, or
if they simply announce that it will become the default in a future release.
http://blogs.sun.com/marchamilton/entry/sun_s_new_modular_blade#comment-1181185661000
This is good thing. I've been using ZFS in place of hardware RAID ( it's faster, due to it's intelligence ) since it was included in Solaris10. Even single drive setups will see benefits ( predictive caching, for example ).
ZFS on a SunFire X4500 ( 48 500GB drives ) just has to be seen.
ZFS on a SunFire X4500 ( 48 500GB drives ) just has to be seen.
You would add a new drive and it would become available as a device. You would then either issue a zpool replace command to swap the new drive for an existing one or a similar add command to add it onto the existing pool. ZFS then formats the drive and restripes existing data while keeping everything accessible.
ZFS is Endian agnostic. If the software that powers ZFS has been ported to the architecture then it will work. Heck, you could export a pool on an Intel mac move the drives to a PowerPC Mac and reimport the pool. ZFS will deal with adjust the endianess of your data on the fly. Of course, your Intel and PowerPC need to use the same hard drive technology, but hopefully you get the picture.
I would think that video work would most benefit from the speed of raided drives (raid z especially).
I don't see why anyone would want to do that.
Here's another good example. Let's say that you have those two drives - a 160 (full) and a 500 (new). You've got an old machine that could benefit from the 160 when you're done with it. I don't know the details on how Apple uses boot disks, but I'm sure that they're easily taken care of; with that in mind, here's the ZFS migration way, using generic terms.
1) Add 500gb drive
2) Pool drives to a single 660gb pool
3) Remove the 160gb drive from the pool
4) Unhook the 160gb drive
Sure, you could do all of the migration yourself, but isn't it nicer to be able to completely detach the concept of a physical disk from a logical storage system? Isn't it very much "the Mac way," to have the system handle those details for you?
And as for the other respondent who wanted full control... well, sure, you can do everything yourself. But the point is that these days the computers are much better about allocating storage than we are. After all, wouldn't you rather be focussing on video production (or whatever) than disk spindles?