photographer-takes-adobe-to-court-for-deleting-photos-worth-250k

In article , Tim Watts
wrote:

The cheaper way is to copy to a decent make of SD card or external SSD
disk (less chance of mechanical failure) but the downside is you really
need to have a yearly regime of checking all devices are readable and an
X-yearly regime of "copy to new device" - even SSDs fail with age.

using sd cards is *not* cheaper, *not* practical and not reliable. a
bad choice all around.

I disagree on all 3 counts. On what basis do you make your arguments?

price, failure rates and capacity.

hard drives are more than an order of magnitude less expensive, more
reliable and available in far higher capacities than is possible with
sd cards (sdxc has a 2tb maximum). sduc can go higher, but that's not
available yet and won't be cheap.

there is also very little infrastructure to support sd cards. have you
ever seen a multi-bay sd card cage? the concept itself is laughable.

are you actually going to swap among multiple sd cards every time you
want to back up??

using an ssd for backup purposes is a waste of money because the speed
advantages are lost. use an ssd for the main drive and spinners for
backup purposes.

You missed the point - spinners are mechanical. They are prone to damage
due to shock (handling and dropping) and if used for offline archival
purposes, run the risk of seizing if left unpowered for long periods
(years).

don't drop them and don't leave them unpowered for long periods of
time. problem solved.

SSDs lose the mechanical problems which greatly increase the reliability.

that's true, but the point *you* missed is that ssds are the wrong
choice for backups. you're paying for speed that won't ever be
realized, with much lower capacity per dollar than with hard drives.

backups do not need to be fast since they happen automatically in the
background. ssds are a complete waste.

But even so, I wouldn't put any device in a drawer and forget about it
for several years, but if I did, I'd bet on the SSD and flash cards
being more likely to still work.

you'd more than likely lose, and there's no point in doing that anyway.

Ideally all files should have a checksum file written with them (MD5,
SHA1 or anything reasonable) and this used to verify files on an annual
basis.

that's automatic with modern file systems.

No it isn't.

it is.

The only common filesystems with *file data* checksums are ZFS, BtrFS -
both linux (and one also Solaris).

in other words, it is.

try to keep your story straight.

also, they're not limited to linux (nor is any file system).

exFAT has only metadata checksumming.

exfat is not a modern file system and is also proprietary. bad choice
all around.

The rest with file data checksumming a SquashFS, ReFS, NILFS and NOVA
and of those, SquashFS is the only one I've seen anywhere in use.

none of those are commonly used.

So yes, you really need to run a checksum generator at the start and
that is the only way you can be reasonably sure your data has not
suffered corruption.

nope.