[OpenIndiana-discuss] experimental or otherwise not recommended options for ZFS

Robbie Crash sardonic.smiles at gmail.com
Mon Jan 30 20:15:30 UTC 2012


If you want to use Fletcher, you need to use verify, as the likelihood of
collisions is increased since Fletcher is not *random*. You don't really
need to verify when using SHA256, and by default, SHA256 is used with
dedup, not Fletcher. . More information about the checksumming and
trade-offs can be found here:
http://blogs.oracle.com/bonwick/entry/zfs_dedup

The bottom line I suppose, is that the ZFS features are stable and
reliable, but may not be extremely useful in your usage scenario.

What Gary said regarding the recompression is accurate if all you're
storing is compressed images, like jpegs. If you're storing raw photo data,
you may see significant savings by using compression.

On Mon, Jan 30, 2012 at 11:01, Jan Owoc <jsowoc at gmail.com> wrote:

> Thank you for the replies.
>
> On Sun, Jan 29, 2012 at 11:33 PM, Sašo Kiselkov <skiselkov.ml at gmail.com>
> wrote:
> > The obvious saying springs to mind: "RAID != backup". If you need your
> > data to be safe, have two copies of it in two geographically separate
> > locations running in two separate machines.
>
> I've read about ECC/non-ECC and RAID != backup and already decided on
> a solution - I wanted some indication of the stability of the ZFS
> features themselves. My question arose when I read that
> "dedup=fletcher4,verify" was causing problems [1] and was removed. I
> was looking for indications whether choosing "checksum=sha256,
> compression=on, dedup=verify" in itself was asking for trouble.
>
> [1]
> http://mail.opensolaris.org/pipermail/zfs-discuss/2009-November/034106.html
>
> On Sun, Jan 29, 2012 at 10:38 PM, Robbie Crash
> <sardonic.smiles at gmail.com> wrote:
> > Those features are thoroughly tested and not experimental. That said:
> > nothing is guaranteed to keep your data safe, and if you're truly worried
> > about it, don't use dedupe.
>
> Because photos tend to be already compressed, and I have only ~5%
> duplicates, I will probably only change "checksum=sha256" from its
> default. Other than CPU time, is there any other reason I may want to
> stick with the default of "checksum=on" (fletcher4)?
>
> Once again, thank you for the insight,
> Jan
>
> _______________________________________________
> OpenIndiana-discuss mailing list
> OpenIndiana-discuss at openindiana.org
> http://openindiana.org/mailman/listinfo/openindiana-discuss
>



-- 
Seconds to the drop, but it seems like hours.

http://www.eff.org/
<http://www.eff.org/>http://creativecommons.org/


More information about the OpenIndiana-discuss mailing list