You've obviously never been in charge of a data archive where, for
example, a company saves its testing scripts, procedures, documents, and
results on hard disks or tapes. Unlike magnets where their effect is
fixed by crystalline structure, dipoles in rust will decay. "Soft"
materials used for recording won't hold their state indefinitely.
Nope, you are right. I never was in charge of data archiving. Although I
do have my own archives that has been around for 30 years now. So I
think that stands for something. ;-)
Floppies degrade faster because of the physical wear presented to them
by the heads actually rubbing against the recording media (rust). They
aren't hermetically sealed so the media and heads get dirty and that
filth damages both media and heads. The media also moves which also
flexes the "soft" recording material (rust).
Yes of course. And with use, I have found that you can boot from a
floppy about 300 to 500 times and then that disk would become useless.
The data sheets I saw, said something more like 800 times. But I think
they were just being wishful on their part. Maybe if you did this within
a week's time, you might get it lasting about 800 times.
That's a different issue regarding corruption of magnetically stored
data. That's where you are concerned about actually changing alignment
of dipoles by influence of a higher magnetic field (like the one used to
record the bits in the first place). I'm talking about dipole stress on
data that is never again rewritten. The loss will still occur with the
tape or disk stored in a vault with shielding around each tape or disk
with humidity and temperature control and without any physical shock
experienced by the media (all of which are other causes for loss of
data).
Yes, but my tests had shown that this is only possible with a
manufacturing defect. Then it can and does happen. And oddly enough, the
Bureau of Standards never saw any of this.
Perhaps you think a decade is longer than anyone wants to save data. As
I recall, actuarial data used by insurance companies must be retained
for a minimum of 50 years. What good would electronically stored data
(on rust) be for geneology information about you if it didn't outlive
you and your decendants? There are writable optical discs claimed to
retain data for 200 years but that's not forever and only if the discs
are stored under ideal conditions in an ideal vault storage space. In a
home or office, the data would survive maybe 100 years for the best
rated discs (only 30 years for consumer-grade media). Even after 100 or
200 years, would you really want all history that was electronically
recorded to fade away after a couple centuries? You really think anyone
would trust history saved on hard disks for more than 30 years, if even
that long? It WILL have to be rewritten (refreshed) to retain that data
(and probably to something other than optical discs by then).
Well I think it is pretty easy to keep data for as long as you want. As
I have data kept for 30 years now.
Rust-based recording media doesn't have anywhere near that longevity for
uncorrupted data retention. Of course, another factor for why hard
disks are not good for long data retention is that them being idle for
so long often results in them siezing. They won't spin when powered up.
When the mechanicals fail, the hard disk is useless and you need a lab
to read from the platters. Removable media can be used by a different
drive when the old drive's mechanicals happen to fail. Hard disks still
have the advantage of higher data volume storage but they'll have to be
refreshed at shorter intervals.
Speaking about hard drives, I've known since virtually day one that any
hard drive can fail at anytime. Most don't though, thank goodness. And
if you want to use hard drives to keep your important data, I would
never trust a single hard drive to do so. I would keep at least two hard
drives to store cloned copies of each other at least. That is what I do.
And the really important stuff I keep on five hard drives at least.
I my own personal experience, my hard drive failure rate is 14.286%
overtime. I have seen two other studies that show their failure rate was
better than mine. Although their's was over a much shorter period of
time. So that and they didn't test consumer grade hard drives like I
did, but commercial grade hard drives which are far more expensive than
what I buy to test.
If you want to talk about other media for storage, well in my experience
tape format is one of the worst (consumer grade anyway). As the tape
tends to stretch and becomes unreadable.
Floppy disks as I mentioned, most are fine overtime while a small
percentage can't even hold data for more than 6 months. I really don't
know how long they can last, but I have tons of them over 30 years old
that I should test one of these days. I did test some of them about 5
years ago and they were still just fine.
As for optical discs like CDs and DVDs. Well those RW formats are the
worst. DON'T STORE ANYTHING WORTH SAVING ON THOSE! As in 5 to 10 years I
don't think there would be much readable on them. The R format are much
better. Although I haven't tested them much either and I don't know how
far you can trust them.
The ROM format is the king as far as longevity in the optical disc
format. I don't think anybody really knows how long they can last. As
you said they claim 200 years, but nobody had one that long yet. And I
don't put too much stock in any of those claims until the real data
actually starts coming in.
Other formats like EEPROMs and flash drives, they say can hold data for
about ten years. You can rewrite them of course for another ten years.
Although I think this figure is very conservative. As I programmed some
EEPROMs myself back in the late 80's that are still perfectly readable
today. The oldest flash drive I have is about 10 years old (without
rewrites) and it too is still readable.
