> Valid point.  I didn't make the clarification that I should
> have. graid3 and gmirror have reached the maturity and
> dedicated to the system, whereas ZFS is still experimental.
> When ZFS is no longer considered experimental, I would expect
> ZFS support in the installer in the same expectation I am
> expecting graid3 and gmirror to be.
> It's all about the status of ZFS itself, rather than the fact
> that it works.

Your point is also valid.  However, our experience with ZFS on the boxes that 
we have installed it has been nothing but positive since about 7.2, and Steve 
Bertrand has also posted that his experiences have been nothing but positive.  
I know that ZFS on FreeBSD hasn't gotten a "stable" rating yet, but it appears 
to be approaching that level and I don't think putting it in the installer (and 
perhaps marking it as "beta") so that more people could test it and give 
feedback about bugs and their experiences would be a bad thing.

To be clear, ZFS itself is indeed stable - our Solaris file servers are running 
it in multi-terabyte configurations on servers that get pounded to the order of 
nearly saturating a 1GB LAN link.  ZFS is the only file system in our 
experience that has suffered no data losses in arrays with more than one 
terabyte (knock on wood).  All other file systems have failed disastrously for 
us in multi-terabyte configurations.  So what you're talking about is not the 
stability of ZFS itself, but the port of ZFS on FreeBSD.

> Does this paint a better picture to you of what I forgot to
> clarify in my original posting?

Yes, clarity is key.  Thanks!  :)

Tim Gustafson
Baskin School of Engineering
UC Santa Cruz
freebsd-questions@freebsd.org mailing list
To unsubscribe, send any mail to "freebsd-questions-unsubscr...@freebsd.org"

Reply via email to