Burned CDs Last 5 years Max -- Use Tape?

Lam1969 writes "Computerworld has interviewed Kurt Gerecke, an IBM storage expert and physicist who claims burned CDs only have a two to five-year lifespan, depending on the quality of the CD. From the article: "The problem is material degradation. Optical discs commonly used for burning, such as CD-R and CD-RW, have a recording surface consisting of a layer of dye that can be modified by heat to store data. The degradation process can result in the data 'shifting' on the surface and thus becoming unreadable to the laser beam." Gerecke recommends magnetic tapes to store pictures, videos and songs."

I guess it depends on how you treat them

[bilbravo]

I have some CDs that are burned copies (although I'd call this great quality cds, not cheap storebrand with no backing), stored in a CD wallet case that are easily over 5 years old... still work great.

Backup media

[morcego]

Anyone who uses anything but real backup media for backup deserves to loose his data.
CDs and DVDs are not for backup. Yes, you can use them to transfer data, or even for short term storage (6 months or less).
HD is also not a good media for backup. If you keep it running, it will break down soon. If you don't, it will also break down, since it doesn't live long without some spinups.
Flash (and other solid state media) also will loose its content in a short time if not refreshed.

We really don't have many options besides tapes. And even tapes are still a problem, since the tape using tend to break down, and you can't find units around for old media (tried to buy a hexabyte unit lately ?).

All in all, tape is the way to go, but make sure you have backups for your tape drives too. Make sure you have humidity and temp control.

I can attest to that...

[ajiva]

The wedding photographer for my wedding gave me a DVD of the video + photos. After about two years the DVDs were so degraded that I could not a single DVD player would recognize them. And that's with light usage... Now I keep important DVD as images on an external hard disk.

Nothing lasts forever

[digitaldc]

When I first asked this question of how long CDs will last, I was told about 70 years.
I was also told that to lengthen a CDs shelf life, always store them vertically in a cool dry place, and clean them from the inside ring to the outer edge in a straight line.

I found an article from the Optical Storage Technology Association and they say it depends on the initial CD quality and handling.
According to this article, unrecorded CDRs last about 5-10 years, manufacturers claim recorded CDRs 50-200 years and recorded CDRWs 20-100 years.

More info: http://www.osta.org/technology/cdqa13.htm [osta.org]

Dutch Study?

[OzPeter]

I can't remember all of the details, but I am sure there was a Dutch group who took a sample of all of the available CDs at the time, burnt data onto them, put them in storage for 2 years and then re-tested the disks quality. Their results showed that all of the disks had significant degredation.

OK .. here is a link to a news report of that study

http://www.cdfreaks.com/news/7751 [cdfreaks.com] This link includes a link to the original Dutch article

To quote:

"The tests showed that a number of CD-Rs had become completely unreadable while others could only be read back partially. Data that was recorded 20 months ago had become unreadable. These included discs of well known and lesser known manufacturers."

 

Re:Museum Archives

[BushCheney08]

This guy [rca.ac.uk] has one. There's even a few pics [rca.ac.uk] linked at the bottom of the page. Perhaps he'll let you borrow it.

NIST Study

[goosman]

http://www.itl.nist.gov/div895/gipwog/StabilityStu dy.pdf [nist.gov]

NIST Did a study that shows up to 30+ years of longevity that is totally dependant on handling and storage.

NIST wrote a research paper on this topic

[jeffmurphy]

NIST has a published research paper [nist.gov] (pdf) discussing how exposure to light and "harsh conditions" affects longevity. NIST also produces a guide for librarians and archivists [nist.gov] (pdf) for the handling and storage of CD/DVD media.

Finally, some have claimed [techbuilder.org] that the glue on the sticky labels might affect the longevity of the dye in the disc, presumably by leaching through the thin top coating of polymer. Search for "glue" in that story, it's half way down or so.

Re:5 years max?

[hackstraw]

I have CDs that have lasted 10 years with no errors. Obviously 5 years is not the maximum life. Perhaps the maximum EXPECTED life.

I've had CDs that were about 5 years old that went bad. They went from the burner to a CD book, and maybe 2 to 5 out of about 100 were bad. I didn't investigate, or maybe even screwed up the burn (win2k), and I used good media, mostly Mitsui.

I believe the tape recommendation to be absurd. If CDs are in the though process, there must not be too much data here. Especially in the context of movies and music. An external harddrive is much cheaper and easier to use than a tape. Depending on your data needs, they can be as little as $80 or so. Drives are easy to navigate and do a restore. Tapes are a PITA.

Now, at work with terabytes of data, tapes in a robot are worth it. Hopefully, I will never have to do a restore, but I have the backup there in case of the need.

My comments are not for "enterprise" type of stuff. More towards a budget minded and low impact if loss data set. I don't consider movies and music that earth shattering if lost.

[C/DVD][+/-][R/RW] != Long Term Archival Storage

[p0rkmaster]

I have always viewed any burnable optical disc as a short-term disposable item - I never seriously considered them as a safe medium for long-term archival storage. Up to the recent past, I would have agreeed with the recommendation that magnetic tape is a better long-term medium....but then I ran into the problem of trying to make an ancient tape drive work. These days, I use good old-fashioned IDE (PATA) hard drives in external IDE-to-USB2/FireWire enclosures for archival storage. The problem of the bearings wearing out doesn't happen if the drive isn't on. As long as it's stored in appropriate environmental conditions - it'll last > 100 years. I have a couple of them that I rotate offsite every two weeks, so one copy is at home and another is at the office. Simple, cheap and very reliable.

Re:Been there...

[Life700MB]

Ok, here's a sample:Vembu's StoreGrid [vembu.com].


--
Superb hosting [tinyurl.com] 20GB Storage, 1_TB_ bandwidth, ssh, $7.95

Re:CD-Rs with a 100 year warranty

[_Shorty-dammit]

when they first gave us CD burners *all* blanks used gold for the reflective layer, this isn't new. They used gold for a reason, but this reason escapes me now. One would guess that it had something to do with gold's resistance to oxidizing. I remember being quite surprised when I first saw blanks that did *not* use gold as the reflective layer, and very quickly avoided them like the plague as they coastered like mad. Yeah, that's a verb, honest.

Re:5 years max?

[lawpoop]

10 years ago, the CDs you were burning were of higher quality, were burned at a slower speed (probably 1x, 2x, or at max 4x), and the burning drive you were using was of a higher quality. You were using high-end equipment and media. Those CDs survived.

I bet that a consumer-grade CD burned last year on a consumer-grade drive purchased last year would not last as long.

Re:5 years max?

[morgan_greywolf]

were cd burners available back in 1995/6?

There have been CD burners available since the late 1980s [roxio.com]

I Archive a lot on DVD/CD, not worried.

[guidryp]

I have never had a failure on one of my archive disks going back to 1997. But I keep these in individual cases stored vertically etc...

Stuff that is not replacable (my personal photos) I burn on two different disk types. I always use high quality disks. Using Fuji TY dvd R+ right now. I believe DVD R disks are a bit more rugged than CD R.

My car disks live in my car in Ottawa Canada. Brutal humid hot summers, I have a set of CD-r in a visor holder. Most of these disks have been in the car for 5 straight years. When I park a disk in my player it often stays for a week at a time. My CD player ejects disks so hot you don't want to touch them. Here I have a few skippers, but each one that skips is also skratched to pieces. Either way. 5 years of torture and most are still fine. I don't think any skip that are not scratched up.

I feel pretty secure about my well cared for indoor disks lasting ten years. Though I will start moving my CD-R backups to DVD.

In ten years, terrabyte storage should be common and cheap.

The correct procedure of using CDs for backup

[hdante]

There's a correct procedure to follow, so that CDs can be used for long time storage. CDs can be successfully used to store data, even though they're so fragile. It's just necessary to do a lot of maintenance. Don't ask me about the details, I'm just writing here what I remember about this.

  1) Burn 2 identical CD-Rs
  2) Store them inside some protected place, upright
  3) When it's necessary to use them, use just CD1. CD2 is never used.
  4) After n years, throw away CD1. (Since the guy said it takes 5 years for the CD to stop working, n should be 2.5 or less, depending on how much you use CD1). Burn CD3. Let CD1 := CD2 and CD2 := CD3.
  5) Goto 2.

  This is a very cheap way to store data.

Re:I can attest to that...

[Firehed]

Working at a video store, I can explain why:

Higher quality drives use, naturally, higher quality DACs. It results in a better video and audio output. However, it also means that they're as a result FAR more sensitive to even the smallest fingerprints and scratches on the disk. Cheap players, on the other hand, use lower quality parts but aren't nearly as finnicky regarding scratches and prints. But the output quality isn't as good. Whenever someone brings back a movie and complains that it skips, we check the bottom (well, we do regardless, but especially then). When it's in good condition, we ask if it's a fairly new/expensive player. Typically, the answer is yes. My $60 Panasonic player has had zero problems, whereas every other player in the house have (mine, for some reason, is the only one that will play region-free disks as well, which is nice as all of mine are. Geek + free rentals... you know)

I don't know if this carries over to CDs and computer hardware, but anything that spits out an analog output (or decoded digital out, as is the case with video DVDs over DVI/HDMI) it seems to be the case. It makes a bit less sense for computer stuff as it's not the drive doing the decoding, it's just spitting out (hopefully) bit-perfect data over the cable, and then the computer deals with that data. I suppose if computer DVD drives were to decrypt movies as they are sent to the computer (which would have been far too consumer-friendly, not requiring them to buy software for their newly-purchased drive so they can watch movies) that would explain things, but when you're dealing with raw digital data, everything should theoretically be identical.

However I've never had any problems that I can recall. However all of my burned DVDs - backup, movie, whatever - seem to be fine to this day. However, I spent $13 and bought 100 DVD cases to keep them in; a big improvement over the binder I had been using, and the spindle (or table/floor!) that some use. For the space-concious, slim cases are available, and almost as cheap. 13c a case for some good piece of mind goes a long way, especially when the disks tend to cost two to three times that.

Re:Screw that - I'm going back to stone tablets

[Fordiman]

Gonna do some math here.

We'll assume that one can reliably retrieve data from a sheet of paper at 200 dpi.

At 200 Dpi, with reasonable page margins of 0.5" per side, you have 1500x2000 (2.86M) potential dots. Assume one bit per dot. That's approximately 0.36MB per page per side. Add one line of dots per side for alignment.

Since a page is evenly divisible by 5000 bytes, lets start there. 75 5000 byte blocks per page; each 5000 bytes will include:
64 bit address (8-bytes)
64 bit CRC (8-bytes)
Data (4984 bytes)

Additionally, since paper is (currently) a read-only media, we can preprocess the data using squashfs, thus assume that 4984 bytes is actually holding approximately 4k to 8k of data after compression and filesystem overhead.

(4k to 8k)*75==(300 to 600kB) per page, per side.

Thus, it would take roughly 175,000 pages, printed both sides, to equal a 200Gb hard drive. At 6ppm, which is pretty standard for a cheap laser printer, that would take 20 days to back up, not accounting for paper jams, toner or sleep.

Re:I guess it depends on how you treat them

[Limecron]

Umm, no it's not.

Data on a CD-R/CD-RW is stored in the chemical substrate which is somewhat clear (usually a green or blue tint). The reflective surface provides the means of reading it by reflecting the laser back through it.

Re:CD-Rs with a 100 year warranty

[Jherek Carnelian]

If they can guarantee 100-year lifespan with a bit of gold then surely the problem with normal discs is over-stated.

The gold is there to replace the aluminum because gold won't oxidize. Kodak used to make similar archival quality discs, I still have a few spindles of them.

Re:I can attest to that...

[Jherek Carnelian]

Higher quality drives use, naturally, higher quality DACs. It results in a better video and audio output. However, it also means that they're as a result FAR more sensitive to even the smallest fingerprints and scratches on the disk.

False. The quality of the DAC has nothing, zero, zilch, to do with the player's ability to read a disc. You might just possibly have had a point if you said ADC instead of DAC, but even then it doesn't make sense - a higher quality ADC will have wider tolerances, not smaller.

but anything that spits out an analog output (or decoded digital out, as is the case with video DVDs over DVI/HDMI)

LOL! There are no DACs in the DVI datapath at all.

Re:5 years max?

[Fordiman]

The drive doesn't 'burn through' the media at all. The dye is what ensures data clarity, and the dye phase changes - meaning it's either all black in that one spot, or its all not-black.

In other words, if six years from now, something you burned with your 'crap' drive isn't working, it's either cos the dis was of poor quality, or because the drive didn't burn the disc properly - in which case, it wouldn't have worked six minutes after you burned it.

Re:I guess it depends on how you treat them

[Yewbert]

Closer than the article, but still missing a crucial distinction. -R and -RW are very different. -R is pretty much as you say - the ink/dye/CHEMICAL layer that's written to is burned; this layer is not the reflective layer.

In -RW media, the write layer is a metallic layer that isn't "burned" but merely heated differentially to create regions of either more-crystalline or more-amorphous metal when cooled; these regions have different refractive characteristics, and can thus be distinguished by laser. This is why it's rewritable - the melting>glassy / melting>crystalline process is reversible.

http://www.usbyte.com/common/Re-writable_CD.htm [usbyte.com]

...contains a succinct but detailed explanation as good as any I've seen (many other sources confuse refractivity with reflectivity, and don't clearly explain that the write layer and the reflective layer in a -RW medium are indeed different layers).

So, to blather on only a little bit longer (too late?), to respond to the immediate contention in this subthread, the reflective layer is in NO case the very same layer as the data is written to. But, in practice, the top coating containing the reflective layer on any -R medium is so bloody thin as to make no difference. If it becomes separated from the surface of the disc, you're hosed. If you want to see how thin this layer is, stick a CD-R in a microwave for a few seconds, till it flashes, and observe the resulting flakes.

Back to the bigger question, the paragraph in this crappy article that says "The problem is material degradation. Optical discs commonly used for burning, such as CD-R and CD-RW, have a recording surface consisting of a layer of dye that can be modified by heat to store data. The degradation process can result in the data "shifting" on the surface and thus becoming unreadable to the laser beam" is needlessliy confusing things by including "CD-RW" in the statement - to conflate a REVERSIBLE phase-change/metal layer-writing process with a PERMANENT burn/dye layer write process is stupid and confusing to anyone who doesn't know better. Whose fault it was to include that, I dunno.

This still leaves the question open as to whether the sorta-stable phase-change alloy ages in substantially the same or else a very different way than the permanently altered -R ink/dye layer, and whether any such difference affects the useful lifespan. I've NEVER seen this specific question rigorously answered. I'd love to hear from anyone who has links or direct info.

Re:Ziplock bags don't breathe,...

[timeOday]

Besides, tapes have a serious format obsolescence problem, because they're unpopular and there are so many incompatible types. Any optical drive you find at BestBuy right now will read a 20 year-old CD just fine, in fact even today CD is the main format for music and shrinkwrapped software distribution. I'll wager that for each of the last 10 years, the number of CDs manufactured outnumbers all the data tapes ever manufactured (and that includes dozens of incompatible types of tape). I think CDs will be readable for a long, long time to come (relative to a human lifetime and not eons).

Re:Just don't drop them!

[WuphonsReach]

This thread also strikes me as funny because I'm in the middle of archiving about 800 VHS tapes to DVD. Many are 15-20 years old, and I've been surprised at how well most of them still work. I wonder if the DVDs will last as long, but I figure it'll be easier to move the data off them since it won't have to be done in real time.

1) Burn 2 copies, store them in physical separate locations.

2) Don't fill the discs to the brim. Only encode about 3.8-4.0GB of MPEG2. Fill the rest of the disc with PAR2 files stored in the VIDEO_TS folder (prefixed with the letter 'z' so they appear on the edges/end of the disc).

I render my DVDs to disc first, add the PAR2 data, then create the ISOs with ImgTool Classic before burning to disc. I make sure that my block size for PAR2 is a multiple of 2048 bytes (CD/DVD sector size).

Even if you can't copy individual files off of the disc, tools like ISO Buster or ddrescue (or dd-rescue) can read the disc back at the sector level. That lets you pull as much information as possible back off of the disc. Assuming you don't have more bad sectors then recovery data, QuickPar (or the open-source commandline tool) can chew on that extracted data and rebuild the files.

I did about 100 VHS tapes a year or two ago. I still have a bunch more to do in the coming year.

Re:I guess it depends on how you treat them

[Jesus_666]

I have been told that DVD-RW is more secure than DVD-R because the Rs dye degrades faster than the alloy used in RWs. DVD-ROM is supposed to be the best choice for backups because like RW it uses metal and additionally it has uses the same error-correction method used in hard drives (bad sectors are marked as such and the data is moved elsewhere. Yes, DVD-RAMs have tracks and sectors like a hard disk and not the single track DVD(-R/RW)s have). Also, according to the specification, DVD-RAMs are supposed to last about thirty years.

Re:CD Presses

[Wavicle]

Gold reflective layer CD-R's with a stabilized pthalocyanine dye have an expected shelf life of 200 years.

That's 4 times longer than the expected life of an aluminum reflective layer pressed CD.

Re:Photography's loss

[Anonymous Coward]

How many people buy DLTtape drives? They aren't cheap and the tapes are not cheap.

I don't know when you last looked at DLT, but you can have a DLT-4000 7-tape *library* for $75, a DLT-7000 single-tape drive for $150, or a DLT-8000 for $350 on eBay. I'll grant you that the new prices are a bit high (in large part because you can't buy new DLT-4000 or DLT-7000) but these things are built like tanks and used DLT should be more than sufficient for home archives. They're also readily available so you could replace the drive if it broke, or even keep a spare if it was really imporant to you.

As for media, a DLT-7000's will write 35 GB on a DLT IV tape -- 4 times the size of a dual-layer DVD, so it should be useful for anyplace that optical media is acceptable. Tapes are not dirt cheap, but retail prices are only about $20/each for DLT IV tapes, which will work for DLT2000+, and you can get new, factory-sealed tapes on eBay for about $10 without much trouble. $0.28/GB for DLT-7000 vs. $0.25/GB for Dual-Layer DVDs.

And finally, tapes will let you write raw TAR archives. I challenge to find a tar archive from anytime in the past 20 years that I can't read with the stanard dd and tar utilities in any modern UNIX system. No requirements for supporting obsolete file systems, no fancy software compression (the drive will do gzip for you), no requirements for fancy hardware drivers, no file name conflicts when @ becomes an invalid character in 2009. It's just a SCSI block device with files in a format you'll can read with tools that will likely be common for longer than you care about whatever is on the media. There's still the problem of being able to read the file contents, but that's not something your archive media can solve.