Richard,

The sub-threads being woven with a couple of other people are very important, 
though they are not my immediate issue.  I really don't think you need us 
debating with *you* about this - I think you could argue our point also.  What 
we need to get across is a perspective.

I am pretty sure that the current Fletcher2 algorithm as implemented does not 
provide the level of security intended by the guy who originally extrapolated 
Fletcher2 from Fletcher for ZFS.  There are no rocks to throw, every single one 
of us, and every system, has goofed once or twice.

The point is that once you experience a few issues of flakey hardware and data 
corruption, then read the ZFS propaganda, you never want to go back.  Yes, 
everything you said was true.  But having seen the vision, we just are not 
interested in being convinced that it is relatively alright.  We have been 
there.  We are interested in a strategy, a roadmap, to move on and get back to 
the vision.

Just remember that we are *here* primarily because we see ZFS as being many 
orders of magnitude more reliable, both in terms of not loosing data, as well 
as telling us when it does.  To dilute this capability is to dilute ZFS' 
differentiation.  To not be transparent is to invite uncertainty and distrust.

Perhaps in hindsight, and given the extreme aversion to risk exhibited by your 
users, the ZFS team might review the checksums used on the zpool level stuff.  
I certainly would, but I am willing to let the ZFS team who understands the 
mathematics, probabilities, and implications better than I make this decision.  
Given that we are very technical customers and don't have our Mac hat on, I 
also believe that it would be appropriate to document some of the rationale.  
There certainly has been alot of material explaining other technical issues - 
why not here?

At any rate, keep on going - we are all behind you 100%.  Please give us an 
open technical solution that we can have 100% confidence in.

--Ray
-- 
This message posted from opensolaris.org
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss

Reply via email to