On Dec 25, 2011, at 10:35 PM, John Stalberg wrote:

> 
> Apple themselfs might not provide this but ZFS from the community could 
> achive what you are asking for I belive?

I think it's a huge hurdle. And that has to do with the Mac OS user base 
itself. They are not experimenters. They are by and large not technical users. 
They are largely unaware of the drawbacks of their own filesystem, if they even 
know what it's called.

Contrast this to the linux open source community, where a large percent of the 
user base is effectively testing new file systems as they are developed.

And a commercialized product for Mac OS X has additional barriers that I think 
will keep its distribution limited. Price is a factor, going concern of the 
company is a big factor. How the implementation is kept up to date in the face 
of constant Mac OS X releases. What does migration look like forwards and 
backwards. But even the premise of why photographers (or anyone) needs a new 
file system needs to be made. By virtue of talking up the superiority of ZFS, 
it inevitably means exposing the deficiencies of jhfs+ to a user base thus far 
not aware of them.

Lots of questions. It's not just about the merits of ZFS itself on paper.

> ZFS is as cool as anyone could wish for I belive, even if it is not of the 
> most reasent version and doesn't have all of the latest bells and whistles as 
> Oracles latest version has?

Btrfs has similar features. While greener than ZFS, it has heavy weights 
putting a lot of development into it including Oracle, Intel, HP, Red Hat, IBM 
and others.

> In your example it seems that OS X unabillity to boot from ZFS today is not 
> the most important issue. It is the 20 TB something large storage that is 
> what needs a top of line fs if I understand what you are saying?

I totally agree. But end users may think differently about this. It's a small 
subset of users who are even aware of filesystems. Most of those with 10+TB 
requirements I've encountered have no idea that their situation is risky, isn't 
best practices for the type or amount of data they have, or that there have 
been solutions to this problem on linux for over a decade. They trust Apple.

On the one hand I'm arguing they shouldn't trust Apple, considering their 
technological storage needs compared to what Apple is offering, compared to 
technologies that exist to solve those needs. On the other hand, I'm arguing 
Apple should do better for a market it has aggressively targeted, which would 
eventually require user trust in Apple to adopt a new storage technology.

> Well ZFS is made to work with large scale storage and look after your data 
> integrity and yet it is easy to control. Wouldn't this make the boot volumes 
> need to be hfs+ a whole lot less of a problem than not having a state of the 
> art fs at all? Perhaps in such a way that your complaint more or less falls? 
> Or is it absolutely nessesary that Apple themselfs makes the FS or ports it?


Absolutely Apple needs to do this themselves, no matter what free or commercial 
alternatives appear. Anyone familiar with filesystems knows Apple is overdue 
for replacing HFS+ of all variants. And for the majority of users who really 
need such a solution, they will want that solution to come from Apple. Do you 
ever think there might be an integrity problem with your filesystem today in 
the course of a Mac OS dot dot release? No. What about even a major release? I 
don't. I wouldn't consider an operating system update would instigate a problem 
at all, because it's a fundamental part of the operating system. A 3rd party 
implemented system will naturally alter this equation. And many people similar 
will not bite.

I think a 3rd party solution is very much an uphill road to reach critical 
mass. But even if it's successful, Apple needs to provide a new filesystem for 
everyone else who isn't willing to be an early adopter, or an adopter of what 
amounts to being a foreign file system.

It's clear Apple have licensing concerns with ZFS. Apple have some GPLv2 stuff 
in Mac OS X.  Btrfs is GPLv2. Apple could get on that bandwagon, if they wanted 
to. It's not GPLv3 like Samba, which was different enough that Apple abandoned 
Samba. But it's still early yet, btrfs isn't done.


Chris Murphy


_______________________________________________
MacOSX-admin mailing list
[email protected]
http://www.omnigroup.com/mailman/listinfo/macosx-admin

Reply via email to