[Tfug] NAS suggestions

[Bexley Hall]

Hi Zack,

--- On Sun, 10/14/12, Zack Williams <zdwzdw at gmail.com> wrote:

[attributions elided]

> >> Larger the better to start with, either 3x3TB or 3x4TB in
> >> Raid5 would be a good starting point.
> >
> > Again, I don't think the need for redundancy/ENHANCED reliability
> > really justifies the added cost/complexity -- I'd rather throw
> > the extra disk space at postponing the day when I've got to sort
> > out what to keep/discard.
> 
> Just an FYI - for large disks, there a not-insignificant
> change of data corruption when doing RAID rebuilds:
> 
> http://queue.acm.org/detail.cfm?id=1670144

Wow!  That was an excellent article!  Yes, I learned how painful
rebuilds/scrubs can become when I built my first 4T array.  It was
that experience that led me to decide *this* should just be a
"hope for the best" solution -- especially since the data is
temporary in nature.

So far, I've just been pushing 10-50G a day at the store, daily,
and building the mirrored copy offline (so my real host isn't
busy, waiting).  That's worked adequately -- even during those
times when I needed to retrieve something from the store *while*
it was updating the mirror.

> > I think I'm going to look at building a small FreeBSD box
> > (I don't think NetBSD supports USB3 reliably, yet) with
> > just basic file services and see how it fares.
> 
> If you do this, I'd recommend using ZFS, as FreeBSD supports
> it well and will checksum all your on-disk data, to at least alert
> you if the disk is going south.
> 
> ZFS's copy-on-write architecture allows you to create on-disk
> snapshots almost instantaneously.

I'll look into it, thanks!  I haven't run FreeBSD since 2.6 so
I've a fair bit of catching up to do!  :-/

--don