deciding UFS vs ZFS

[Paul Kraus]

On Jul 22, 2014, at 8:33, RW <rwmaillists at googlemail.com> wrote:

> Would you rather lose a third of your books, or a third of the
> chapters from all your books? 

If you are storing data you do not want to lose on non-redundant storage, well, then you deserve what you get. Whether it is UFS or ZFS.

What I really do not understand is this mindset that using ZFS (on a single drive) is an all-or-nothing proposition in terms of failures. What kind of failure with a partitioned drive and UFS will yield *less* data loss than ZFS? Bad disk blocks? ZFS sees the bad checksum and lets you know. What does FreeBSD UFS do with bad blocks (or silently corrupt blocks)?

On the other hand, having to guess up front how much space will be needed in each of the various (manually managed) partitions is a crap-shoot. More often than not leaving lots of unused space that *other* partitions could really make good use of. That is, in my opinion, the biggest management advantage of ZFS on a single drive… pooled storage with the ability to control it (quotas and reservations).

But I have only ever run ZFS on a single drive for testing purposes. All of my real data is on redundant storage. I rsync the data directories on my laptop with my server at home on a regular basis (and even use Time Machine, yes the laptop is a Mac) at the office, so I have THREE copies of the important data (one of which is redundant). The cost of storage and low end servers is much, much less than the cost of lost data.

--
Paul Kraus
paul at kraus-haus.org