Article re. CD durability

[john v.]

<p>For those of you who still archive images (or anything else) on CD's, here's an <a href="http://www.techarp.com/showarticle.aspx?artno=641">article</a> that refers to some actual results regarding CD's that were about 9 years old and how many of them were still readable after all that time.</p>

[William Kahn]

<p>Interesting, but basically anecdotal. A really valid test would need controls on the recording process, the media and the method of testing data integrity over time. It might be helpful to remember that a 7% error rate is also a 93% success rate. At any rate, being paranoid, I always back up on both EHDs and DVDs......</p>

[brad_smith8]

<p>Often overlooked are the chemicals in felt pens commonly used to label CD's/DVD's. They can break down the plastics and reflective layers that the laser needs to bounce off of. So it's important to use pens specifically made for that purpose.</p>

<p>And of course everything on the planet (indeed the universe) will degrade over time. The best we can do is only to slow it down.</p>

<p>Back in the '80's a wise computer sage said to me, "There is no ownership of data unless you have two <em>verified</em> copies at home, and a third copy stored at another location (in case of a fire, etc.)" Back then we used tapes, but they all had to be continually verified.</p>

[leicaglow]

<p>Even 10% is not acceptable to me. As for my experience with 9 year old CDs, my failure rate is far higher than 10%. I estimate in the 80% range. I have been recovering some of them with software that yanks the files even though the disc index is corrupted, but still, these are disks with 100 year ratings. I think someone should be held accountable for false advertising.</p>

[keith_b1]

<p>It seems clear at this point in history that CD/DVDs cannot be trusted long-term. Storing them properly helps. Properly = in the dark. Not in a jewel case on the bookshelf in a room in the house. Total darkness. My audio cd copies that I recorded in 2003(on Sony and other brand consumer CD-Rs), and stored in black, more or less light tight Case Logic slip-in cases <em>in my car </em>still play fine, but I have no idea for how long.</p>

[frank_dzambic]

<p>My personal opinion is that the number one factor in disc longevity is the burn speed used to record them. I can count on one hand the number of discs I've had fail on me and I know I burned them at the highest speed possible at the time. All the rest were burned at a maximum of 4X, dvd and cd, and I've yet to have one of those fail years later. It takes a bit longer to burn them, but not nearly as long as trying to recover the data on a failed disc. They're Verbatim Data Life Plus discs if anyone's interested, but I still think the recording speed is a bigger factor than the brand name though I do trust the brand name as well. Why should burn speed matter? I don't know. I wonder if differences in discs burned at various speeds could be seen under a microscope?</p>

[Tim_Lookingbill]

<p>Sure wish that article would've been more thorough. I've done lens reviews that were more informative. Why are folks who probably get paid to write such articles on the web so darn lazy?</p>

<p>Other than that I have to agree with the premise of what they're saying. Here's what I discovered about CD-R's five months ago...</p>

<p>http://www.photo.net/digital-darkroom-forum/00SO0y</p>

[john_deerfield]

<p>Burn speed matters because of data <em>integrity</em>. Burning a disc more like a bleaching process than a burning process where the dye layer is "discolored" resulting in a bump or pit to be read as a 0 or 1 (way over simplified). Those that argue that burn speed doesn't matter since it's just data, just 1's and 0's don't factor into account the ability of the reader to interpret the data. A slower burn gives a "better" bump or pit that is more likely to be read by any number of readers or in the case of archiving data, to be read by a reader at some point in the future. The catch now-a-days is not to burn too slow. Burning too slow on dyes that are made for faster burners can result in data integrity being harder to read by other readers. I usually suggest a middle of the road speed for burning. Often, when you burn a disc it gives you several burn speed choices, I don't choose the slowest, but I don't choose the fastest either. But when in doubt, I always go slower. For what it's worth, Blu Ray media is suppose to be designed with archival quality in mind. As far as buying into claims that CDs will last 100 years or more... let this be a lesson learned. When such claims had been made CDs hadn't been around for 100 years so how could anyone <em>know</em>.</p>

<p>Lastly, if you want another burn speed analogy, look at mini-DV tape (or if you want to go analog, VHS tape or audio tape). DV tape is digitally stored information. However, it's recommended to record in SP mode NOT LP mode. You have less recording time in SP mode but less change of drop out or other corruption of your data. </p>

 

[Ed_Ingold]

<p><em>When such claims had been made CDs hadn't been around for 100 years so how could anyone know.</em></p>

<p>The projected life is based on accelerated testing, which in turn is based on <em>Arrhenius' Law</em> (q.v.). This is accepted practice in the scientific and engineering community. It's no big deal to write at 16x speed. If you can't get a good burn at that speed, it won't help to use a lower speed. In fact, you usually get more errors at 2x or 4x than at the rated speed.</p>

<p>Unless discs have been subjected to abnormal conditions or physically damaged, problems can be attributed to bad burns. These discs were bad from the get-go. At a minimum. all discs should be compared to the original files ("verified") upon completion. The false acceptance rate is fairly high for verification, so a more comprehensive readibility test should be performed on a sampling, and if you incur any verification errors. I've written on this subject in more detail in the recent past.</p>

<p><em> However, it's recommended to record in SP mode NOT LP mode. You have less recording time in SP mode but less change of drop out or other corruption of your data. </em></p>

<p>This is not a good analogy. VHS or digital tape is recorded at a higher density or lower bandwidth in LP vs SP mode. The data density on a CD/DVD is the same regardless of the speed.</p>

<p> </p>

[john_deerfield]

<p>Say what you will. I have literally burned 1000's upon 1000's of DVDs, probably more like tens of thousands and DVDs burned at a slower speed have a higher success rate across the board. Again, it's not whether the pit or bump is a one or zero, it's how well <em>defined</em> the pit or bump is that allows another laster to read data. While I didn't understand it at the time, the first time we ever burned a dance recital video to DVD I must have had 30% of the discs returned. These were verified discs. The only thing I changed (after extensive web research) was to use a slower burn speed. None come back after that. Beyond my experience lies common sense. When things are done "faster" there is more likelihood of error or at least unexpected results. Finally, not to diminish the contributions of science or accepted practice, but often that is simply our best guess given current information. Heck, Viagra was a complete accident! In any event, while I respect the science (my son is a U of I engineering student!), I must evaluate the accepted science with my own experience. </p>

[Ed_Ingold]

<p>If you have Plextor drives, "Plextools" has a powerful set of diagnostic tools. These are much more sensitive and parametric than the verify after burning function. Infinadyne makes diagnostic and recovery software which works with all drives, which measures the frequency of soft and hard errors.</p>

<p>Files which fail verification are invariably bad, but files which pass are not necessarily good. The computer may try to read a sector half a dozen times or more and pass the first time it succeeds. A disc with 300 errors/sec will pass verification most of the time. A "good" disc has about 5 errors/sec. In statistical terms, "verification" is not very robust.</p>

<p>Not all discs are equal, and the quality for a given manufacturer may slip. I have consistently good results with MAM and Taiyo Yuden CDRs and Verbatim DVDRs (+ and -). I have consistent problems with Taiyo Yuden and Memorex DVDRs. Next week, who knows?</p>

<p>Best guesses based on inadequate knowledge are usually incorrect. That's why people believe in ghosts and "universal" health care, among other things.</p>

[john_deerfield]

<p>I agree with all of that. But might add that all knowledge is inadequate. I also agree with Frank that burning at a slower speed decreases your chances of having a disc that another laser can't read. I believe this because I have experienced it. You don't have a 30% return rate drop to 0% with the only factor changing being the burn speed and then look for some other reason? While verification may not be very robust, it's what is readily available and when pumping out 500 discs of something I doubt I would have time to run each burn through another diagnostic. If a good disc only had about 5 errors and a bad disc 300, whose to say that I am not reducing the errors simply by using a slower burn speed? For me it's just a simple solution: slower burning results in a better disc. I am not going to boot to Windows and run a diagnostic on every burned disc. Using my methodology, the <em>only</em> problem I have encountered in the past 5-years was a video DVD that simply didn't like a variable bit-rate. The material was re-encoded at a constant bit-rate and all was well.</p>

[Ed_Ingold]

<p><em>If a good disc only had about 5 errors and a bad disc 300, whose to say that I am not reducing the errors simply by using a slower burn speed?</em></p>

<p>Simple! Run the diagnostics on discs burned at 8x and 2x. See if there is any difference. When there is a real problem, like bad media with 300+ errors/second, the burn speed has no effect. With good 16x media, you get about twice the error rate burned at 2x than at 8x. However, the error rate is so low that it doesn't matter, except that it takes a lot longer to burn a batch of discs.</p>

<p>8x is at the bottom end of the CAV (Constant Angular Velocity) range for DVDs. 6x is usually the top end of the CLV (Constant Linear Velocity) range, which gives a burn with less jitter. Either works well enough for production. I suggest closer monitoring for higher speeds.</p>

<p>You don't have to run the diagnostic on every disc, but it pays to do it at the start of a big run, or at least before you release it to the public. It also pays when you change brand or type of disc. That's why I used the term "sampling". If you would like to see a good sampling plan, look up Mil-Spec 105D (or E) - obsolete but statistically valid.</p>

<p>A reject rate of 30% is absolutely toxic! Even 0.3% would be bad for business. If that's the best you can do, farm it out to someone who knows what they're doing.</p>

[Ed_Ingold]

<p><em> the only problem I have encountered in the past 5-years was a video DVD that simply didn't like a variable bit-rate. The material was re-encoded at a constant bit-rate and all was well.</em></p>

<p>VBR vs CBR is the method of compressing the video prior to compilation. Some DVD compilers have specific requirements (e.g., Pinnacle), others take practically anything (e.g., Adobe Encore or Final Cut). It has absolutely nothing to do with the way the disc is burned - digital words are digital words.</p>

[john_deerfield]

<p>CBR uses well, CBR. There are still rate spikes in the encoding but they are not as drastic as what there can be using VBR. Some DVD players do not like drastic changes in the bit-rate. Yes, this has nothing to do with slower burn speed other than to enforce the idea that I have been using slower burn speeds and have never had a disc issue OTHER than my CBR/VBR illustration.</p>