Major news of the week in the geek world is that OSX Leopard is going to use the open source filesystem from Sun Microsystems, ZFS, as its default. Why use ZFS? Well, for a start it has no theoretical maximum size for a single file, unlike say NTFS which is pants. It also has storage polling which means that you don't need to create volumes when you add for drives, it works a little bit like virtual memory in that sense.
There is as well the Z that matters too. It stands for Zetabyte. This is because whilst it has no theoretical maximum size for a single file, the filesystem itself has a ceiling limit of a zetabyte. That's approximately one million terrabytes, or all the porn in the world and then some.
It also has support for intelligent incremental snapshots. That means if you edit a file it will only snapshot the edits and use the original as a referential point reducing the need for massive disk loss like that which you get with say "System Restore" in Windows. Oh yes, and it really is very very fast.
Apple are making a wise move going down this route. As we see continue to a growth in storage size coupled with a reduction in the physical space it takes up, ZFS becomes the most expedient and sensible option. Microsoft have a lot of work to do on this matter and should really be thinking about dumping NTFS once and for all.
Note: Apologies to readers who have no idea what I am talking about.
10 comments:
So what your saying is a file can cross many hard drives and you still think it's on one volume ,what about if a drive trashes ,can the file still be recovered or will just that part of the file that will be lost or could the file be totally lost.
Having no idea what you're talking about doesn't dimish the enjoyment of seeing you wax lyrical.
Is it new, or have Sun just been keeping it to themselves as they didn't know what to do with it?
No it's not new, it's been around for a while.
First Anonymous: what you are describing there is a Raid system (or one version of it). I don't believe that ZFS works like that (but I'm sure Dizzy could enlighten me).
Diz, I knew about this some time ago actually: here's an article from December 2006.
DK
Yes DK< that is RAID that he's describing, but that post you link to is slightly different. It's now being confirmed that ZFS will not just be supported but be the defaul filesystem type.
This will come in useful. The amount of data that the nuLieBore government want to keep on us is going to need this.
How does it behave fragmentation-wise?
I don't have any particular gripes with NTFS (not that I adore it, or anything) apart from the pain of defragging really large volumes.
This guy seems to have a good handle on ZFS, much to the chagrin of astroturfing Sun employees!
[Search blog for "ZFS"].
Further to the above it seems that ZFS encapsulates file system, volume management and RAID functionality - eliminating the need for expensive RAID controllers.
Apparently you can even use a bunch of cheap disks instead of a dedicated array.
Post a Comment