David Madore's WebLog: CD and DVD longevity

Index of all entries / Index de toutes les entréesXML (RSS 1.0) • Recent comments / Commentaires récents

Entry #0623 [older|newer] / Entrée #0623 [précédente|suivante]:

(Saturday)

CD and DVD longevity

Following a recent discussion with some friends, I have tried to find some accurate information (on the Web or elsewhere) on the stability (longevity) of recordable optical media (CD-R[W] and DVD±R[W]). The bottom line is this: nobody has a clue; it is more or less agreed that under ideal storage conditions (away from dust and humidity, at a mild room temperature, and in complete darkness), most CD's should remain readable for a century (this is a mere order of magnitude, not an accurate assessment), and that when subject to “typical” handling and use (or not-too-careful storage) they only last for a couple of years. This is such a gross estimate, however, that it might just as well be summarized as nobody knows.

Many factors come into account, and the disk can degrade in many ways and in various places. It used to be thought (rightly or not) that the CD-R organic dye, which by definition is highly photosensitive, was the most vulnerable spot, and fruitless debates have been formed over the alleged superiority of the golden (phthalocyanine) dye over the green (cyanine) or blue (metal AZO) ones; but over times, manufacturers have seemingly been able to extend their dyes' life spans by adding various stabilizers. I am told that CD-RW's are more stable than CD-R's in this respect, because the recording layer alloy is not photosensitive: recording is achieved by heating the material either to its melting point (so that it cools down rapidly in an amorphous state) or to an intermediate crystallization point, according as a “pit” or “land” is desired (the amorphous and crystalline phases have different optical properties), and the crystallization point is sufficiently high that one can be confident that the material will not spontaneously return from its metastable amorphous state to stable crystalline state, at room temperature, within a time span of decades. Still, I am not too tempted to advise people to record their important backup data on CD-RW's, because the disk has other weaknesses than just its recording layer (whether optical dye or metallic alloy), and the CD-RW's lower inherent reflectance (even in crystalline state) makes it more vulnerable to certain other degradations (for example, in the substrate). Incidentally, (pressed) CD-ROM's are, of course, not photosensitive at all, but they are also not eternal.

Other possible causes of disk damage include: dust, which is abrasive and can both damage the substrate and penetrate the very thin (and hence easily damageable) protective layer to the reflective part; or water, which is absorbed by various organic compounds (such as the polycarbonate substrate). Or just about any kind of handling, which can slightly deform the disk, or scratch it (the latter being, naturally, the worse one). Ink, of any kind whatsoever, is not recommended (either the pen will scratch the disk, or the ink's solvent will penetrate through the protective layer and might damage the reflective layer): if one needs to label one's disks, one is better advised to take note of the medium's serial number (generally printed on the inner surface) and associate the label with that number, rather than writing it actually on the disk.

This page on CD-R longevity also says a couple of interesting things about the question, and points to a study attempting to measure CD-R longevity, which is somewhat inconclusive but nevertheless interesting. (For example, one learns that errors are much worse if the disk is aged before recording than before: so the moral of that is that one should always record on fresh new disks insofar as possible.)

It would be nice to have access to the number of errors detected and corrected at each level of error correction. (Called C1 and C2 on a CD: see this page about compact disc errors for a summary of how they work. Those who wish to know more on error detection and correction mechanisms are advised to read the ECMA 267 standard for the DVD, which specifies them in detail: although the CD-ROM is an industry proprietary standard, the DVD standard is a freely available ECMA publication, and I believe the error detection and correction systems are actually identical between the two.) I'm not sure whether the drive can be forced to report errors below E32 (uncorrectable after second level of parity) to the operating system, but it would be useful, especially for E22 errors (last correction between E32), which are a sure sign that the disk is just about to become unusable.

Another aspect of longevity, which is too often overlooked, is drive longevity. In my experience, a CD drive in a computer will give definite signs of old age after a couple of years already, and will be mostly unusable after a mere five years. The main culprit here is—without any doubt—dust: I have terrible dust problems in my computers, and fans and CD drives (I have completely abandoned floppy drives a long time ago) are the parts which suffer most from it. The sure sign that a drive is aging is that it frequently finds errors (or sometimes just takes a long time to read, and slows the rotation speed to minimal, because it is trying over and over again) in disks which other drives can read without problem. Given how cheap drives have become nowadays, this isn't so much of a problem now, but someday CD's will have become an obsolete technology, and drives won't be made any more (or they will have been crippled by DRM schemes and whatnots), and one will be easily stuck with gigabytes of data on CD-R's that are still readable in principle, but not in practice for lack of a working drive!

↑Entry #0623 [older|newer] / ↑Entrée #0623 [précédente|suivante]

Recent entries / Entrées récentesIndex of all entries / Index de toutes les entrées