In message
<0ce5fa3e-a630-4620-9c96-036b68edc82d@o10g2000vbg.googlegroups.com>
someone claiming to be Castor Nageur said:
QuickPAR can recover up to 10% of any piece of data located anywhere
of your corrupted file.
10% isn't a hard number either, you can recover up to whatever
percentage of data you want based on the amount of PAR2 blocks you
created in the first place.
Nothing stops you from creating 100% PAR2 blocks and throwing away the
entire original media.
Also, even if you set the threshold at 10%, you can't necessarily throw
away 10% of the original media. You can throw away or damage 10% of the
blocks of the original media, but you couldn't damage 10% of every block
and recover.
This difference is mostly academic since damage will often be grouped
together, but it's worth considering if your fear is long-term bitrot.
Imagine, you saved an archive on a DVD, forgot it and stored it in bad
conditions.
If you retrieve it damaged years later, you will still be able to
recover your data (if no more than 10% of your data was damaged).
Obviously, I can not imagine generating some QuickPAR redundancy files
on my 100 GB Acronis backups because it would take an entire day. The
idea was Acronis could do it on the fly (like WinRAR does) but I do
not think it does it.
I'm not sure why your imagination precludes letting a computer run for
several hours at once, but it's entirely possible, it just takes a
while.
QuickPAR takes a long time because the math is complex (it's CPU-bound),
and the fact that it runs after the fact vs during the original backup
wouldn't make much difference in terms of the total runtime. There are
other algorithms that might be more efficient on CPU resources, but few
that offer the redundancy that PAR2 does.