Frankly, while I love Linux for a bunch of things (it runs my desktop, 
amongst other things), I really think the impetus of "Geeze, can't we 
get Linux to have THAT" every time some interesting thing comes along is 
a BAD IDEA.

Having a large ecosystem of OSes which can cross-germinate ideas, but 
not necessarily code, is a Good Thing. However, trying to make one OS do 
everything well is a recipe for disaster, let alone code bloat.  It's 
like trying to design a car that: (a) does 0-60mph under 3 seconds (b) 
can haul 2 tons of cargo (c) get 50mpg and (d) cost under $15k.

Not going to happen, and the compromises that you have to make to 
achieve certain goals make other attributes either difficult or 
impossible to accomplish.   OSes should have a "optimum" targeted user 
population - one that  gain the most benefit from the OSes' strengths, 
and are hurt least by it's weaknesses.   Trying for a One True OS is a 
silly (and dangerously shortsighted) goal.

In ZFS's case, if the ideas around it are so attractive, then 
reimplementing the ideas in a Linux-compatible FS module are entirely 
possible, even if certain patents must be worked around. I'm not 100% 
sure, and I'm certainly not one of the lawyers, but my reading of the 
ZFS internals and related IP leaves me with the impression that a 
complete clean-room reimplementation is entirely possible, though it may 
not have the same performance profile.  That is, the ZFS on-disk format 
isn't IP protected, and the general concepts of how ZFS works (pools, 
CoW, snapshots, etc) are open, it's just _how_ the guts do these things 
which are.

So, I'm actually pretty glad that ZFS will remain solely in the 
CDDL-compatible land, and that if a ZFS-compatible system shows up in 
GPL-land, it will have a different underlying engineering, which may 
provide many more interesting insights into improving what is truly the 
FS-for-the-new-millenium.


(Please, if I'm wrong about our [Sun's] patent protection of ZFS's 
internals, I want to know _now_. Speak up and correct me please, folks).

-- 
Erik Trimble
Java System Support
Mailstop:  usca22-123
Phone:  x17195
Santa Clara, CA
Timezone: US/Pacific (GMT-0800)

_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss

Reply via email to