Data archival for the circumspect [2025-05-13]

Unfortunately, there seems to be no perfect medium for information. Paper cellulose breaks down, quipus fray and unravel, hard drives break, and flash storage goes bad. The Sumerians demonstrated you can print data onto a clay tablet and it will last for thousands of years, but unfortunately with how much digital garbage we keep today, this is would be expensive, time consuming, and take up a lot of space (please do note that even the clay tablet is not safe due to proton decay). The question then is of a trade-off between longevity and cost.

What medium?

In the matters of cold storage, flash memory (this includes USBs, SD cards, and SSDs) has the worst data lifespan due to charge loss. Anecdotally, more flash media has failed or been corrupted than any other kind for me. It should not be used for cold storage under any circumstances.

Magnetic storage provides a better avenue for cold storage, though the mode matters. An HDD is preferable to an SSD, though LTO is much greater than an HDD. LTO provides some of the best cold storage, and is extremely cost effective for size. The main issue is the cost of a drive, which is thousands of USD. However, if you can afford such a cost and have need to store more than a few TB of data (if it's less, just use HDDs and replace them every few years), it is a close to ideal option.

Finally, optical media (CD, DVD, Blu-ray) is what I propose to the average person; particularly BD-Rs. As opposed to CD-Rs and DVD-Rs which use an organic dye, BD-Rs utilize an inorganic dye which is much less susceptible to decay. BD-Rs are less cost-effective by storage than LTO, but do not require a multiple-thousand dollar drive to operate (though don't skimp out and buy a BD cheap drive—you'll regret it! *maybe). They also have a great lifespan. Ideally we would used pressed discs since those seem to last ad infinitum, though I doubt many interested have the capital for a mastering machine, injection moulding machine, and so on. For these reasons, I will focus on utilizing BD-Rs for archival. As an aside, I do not recommend using rewritable/recordable erasable discs (CD-RW, DVD-RW, BD-RE) for archival because of the degradation of their phase-change materials, as well as their often lower drive compatibility. Also, some newer BD-Rs (LTH) apparently use organic dye (instead of HTL), avoid those for the reasons just discussed in this paragraph.

Note: most importantly, don't store your data somewhere hot, humid, or stupid.

Burning

In order to burn the files, they should first be combined into an ISO. We can use mkisofs for this purpose. Place the files you wish to burn to the disc and use mkisofs to create the ISO we will burn.

mkisofs -o archive.iso -V "ARCHIVE" -UDF -iso-level 3 bdarchive

You may prefer to add Joliet, Rock Ridge, etc. Remember to review the man page for any other options you may wish to or not to include. Before burning, I recommend you also utilize mkisofs' -print-size option to determine whether or not the ISO you are about to create will be too large for the disc.

Ensure your disc drive with disc is connected to your machine. Find your device with -scanbus, and burn your ISO. These are the options I typically use:

cdrecord dev=x,x,x speed=2 driveropts=burnfree -dao -v archive.iso

Choose the slowest possible speed to reduce write errors and stress, and again, remember to select options as relevant to your disk and drive.

After burning, mount the disc and make sure everything burned correctly.

sudo mount /dev/sr0 /mnt
diff -r /mnt bdarchive

If any discrepancies are found, make sure to look back over cdrecord's output. A variety of factors can cause differences, including a bad disc, bad drive, or random chance. This is why I earlier recommended not to cheap out when buying a drive. Old drives will also need to be replaced over time. Before burning for archival, I prefer to burn a cheaper test disc to check early for any potential failures, and avoid wasting more expensive discs and time. Once you've finished burning, all that's left is to properly store your discs. Keep them away from humidity, light, etc. Generally they should be stored back in whatever jewel case or spindle they came with, unless it's something cheap. You should probably label them, too. I use a Dewey Decimal-inspired system, along with printed records (should I use archival-grade paper and ink? Deeper down the rabbit hole we go...)

Feel free to contact if you have any further inquiries.