Optical Disc Life Expectancy: A Field Report

Optical Disc Life Expectancy: A Field Report Barry M. Lunt,a Douglas Hansen,b Matthew R. Linfordc a Information Technology, School of Technology, Brigham Young University, Provo, UT 84602; bMillenniata, Inc., Lehi, Utah 84043; cDepartment of Chemistry and Biochemistry, Brigham Young University, Provo, UT 84602

Abstract The lifetime expectancy (LE) of optical discs has often been determined with accelerated testing. This paper reports on the LE of discs subjected only to normal controlled conditions of temperature, humidity and light.

Introduction It has been argued that there is an urgent need for a permanent method of storing the world’s digital information,1,2 and there are published studies that have described the LE of recordable optical media and magnetic tape media.3,4,5,6,7 The outlook was not too good: depending on which report you favor, the LE for recordable optical media was from 7-25 years, and that for magnetic tape from 30-50 years – and this is the average LE, not the minimum. But what users really care about is not the average LE, but the minimum – when the first data loss will occur. None of the documents cited above have reported on this. In addition, data can be stored on magnetic hard-disc drives (HDDs) or on flash drives, also known as “jump drives”, “USB drives”, “thumb drives”, etc. However, the relatively short LE of HDDs is well known, as well as their tendency to fail catastrophically. As for flash drives, the LE for the data, using this technology, has been quite well known since the invention of its predecessor, EEPROM, in the 1970s, and it is only 10-12 years. The purpose of this paper is to provide an update on the LE for optical discs, as they appear to have the greatest promise for a true archival-quality method for preserving digital data. The references by Byers, Iraci et. al, Navale, Shahani et. al, and Slattery et. al, are becoming dated (2004-2005), and they did not look specifically at the very best recordable optical media available. So it is time for an update.

Recent Accelerated Aging Results Most manufacturers of recordable optical media do not advertise the LE of their media. The notable exception is in archival-quality media, where CDs are commonly advertised to have an LE of 300 years, DVDs an LE of 100 years, and BDs an LE of 200 years. It is very difficult and time-consuming to perform LE tests, so many times media are tested by simply comparing their degradation of digital errors to other existing discs. In a test reported by Svrcek8, five brands of archival-quality DVDs and one brand of a top-rated standard-quality DVD were subjected to accelerated testing under conditions of 75°C, 85% RH, and 1120 W/m² full-spectrum light (the equivalent of the normal intensity of the unclouded sun at the equator at sea level). Failure was defined as exceeding the PI8(max) specification of 280, as tested with the ShuttlePlex Media Analyzer from Media Technics. Table 1 gives the initial and final (after 24 hours of testing) values for PI8(max) for each of these DVD brands. From this it can be seen that five of these brands failed after 24 hours of testing, while one brand was apparently unaffected by the testing. Table 1: Initial and final PI8(max) values for five archival-quality and one top rated standard-quality DVD.

PI8(max) Initial 24 Hours

Brand 1 18 >5000

Brand 2 32 >5000

Brand 3 34 4249

Brand 4 102 3284

Brand 5 110 2096

Brand 6 122 83

It should be pointed out that the accelerated testing conditions to which these discs were subjected were not as severe as can be encountered on the dashboard of an automobile left in the summer sun in a humid climate. Consumers commonly leave their recorded optical discs in such environments, yet expect them to retain the data. Based on the above test results, it is highly doubtful that the data on five of these recordable DVD brands would survive such treatment.

Recent Normal Aging Results: Field Data on Two Collections The authors have been able to obtain data on the recordable optical disc collections of two university libraries. These libraries have chosen to remain anonymous, due to the sensitivity of the information in their possession and their desires to retain it indefinitely. Library #1 has a collection of approximately 8,500 DVDs, all recorded since 2003. These discs are of eight different brands, including both standard quality and archival quality. These discs are non-circulating discs, meaning that they have constantly been stored under conditions of controlled temperature, humidity and light exposure. These discs were accessed in 2010, and 177 of them had readback problems ranging from unreadable files to unreadable discs. So, in a maximum of 7 years of natural, controlled aging, 2.1% of their collection had experienced permanent data loss. Library #2 has a collection of approximately 18,000 optical discs, all archival quality, and nearly all of them CDs (about 99.5%). It should be noted that the expected LE for CDs is longer than for DVDs. As with library #1, these discs were non-circulating discs, so they never experienced storage conditions other than controlled temperature, humidity and light. They ranged from 4-14 years old when they were tested in 2008/9. Of this collection of discs, 319 had either a permanent loss of data or they were completely unreadable. So, in another example of discs kept at conditions of controlled aging, in 4-14 years 1.79% had experienced permanent data loss.

Analysis Twelve of the failed discs from library #2 were obtained for further analysis. These were analyzed on the ShuttlePlex Media Analyzer from Media Technics. Their maximum digital errors are reported in Table 2. Table 2: Maximum digital errors on 12 discs from library #2. CD # 1 2 3 4 5 6 7 8 9 10 11 12 Avg StdDev

BLER 5783 226 6522 1984 7160 238 285 42 46 1748 100 970 2092.0 2744.6

E-11 1078 99 1235 1532 1275 167 223 32 28 1299 99 407 622.83 599.77

E-21 272 30 531 431 1156 46 61 8 10 365 10 275 266.25 336.19

E-31 5423 123 5941 115 5381 38 69 21 30 87 51 544 1485.3 2477.7

BERL 1171 17 1209 6 1034 5 9 2 3 2 7 84 295.75 509.90

E-12 438 883 310 338 952 344 677 242 415 340 448 512 491.58 228.48

E-22 297 837 203 384 182 296 704 13 27 57 321 380 308.42 252.97

Uncorr 5730 79 6318 48 6072 5 6 56 56 4 96 720 1599.2 2687.9

It is obvious from the values in Table 2 that these discs (all archival-quality CDs) are indeed well beyond failing. The error specification calls for a maximum value of 220 for BLER(max), and a maximum value of 0 for Uncorrectable Errors. All but 3 of these discs exceeded the 220 value for BLER(max), and all of them exceeded the 0 value for Uncorrectable Errors. Some of these far exceeded the maximum recommended value. And while there are no standard recommendations for the maximum on the other error types, it is obvious that these discs have failed. Standard-quality CDs have an LE of approximately 25 years; for archival-quality CDs, the LE is advertised to be 300 years. For standard-quality DVDs, the LE is about 10 years; for archival-quality DVDs, the LE is advertised to be 100 years. Using these values for the average LE for each type of disc, we can estimate a distribution for the LE for each disc type.

First, we assume a normal distribution in the LE for all of the discs in the collections of these libraries. Next we assume the average lifetime of the discs from library #1 to be 7 years (which is very generous, considering that some of them surely failed much sooner than that). Then we take the average LE of these discs to be 100 years. Using these assumptions, we can calculate that the standard deviation would be about 46 years. This means that the consumer has basically no idea when their first data failure will actually occur if their data is stored on these discs. Doing the same for the CD collection of library #2, assuming the average LE to be 300 years (they were primarily archival-quality CDs), assuming the average LE to be 10 years, and knowing the failure rate to be about 1.79% at 10 years, we can calculate the standard deviation to be about 138 years. Again, these results indicate that the consumer has basically no idea when to expect the failure of their data. Curators from these institutions have reported that they have observed some discs to fail in as little as two years, which can be expected given the standard deviations calculated above.

The Important Parameter Our research leads us to argue that the most important parameter to consider when choosing optical disc media for long-term data storage is not the average LE of the media. The previous section has shown that even CDs with an average LE of 300 years, or DVDs with an average LE of 100 years, can be expected to have a failure rate as high as 2% in only a few years. We emphasize that it is not the average LE that consumers really care about, but that is the only parameter that is advertised or available. The most important parameter to consider when choosing optical disc media for long-term storage is the minimum LE for the discs – but this is never reported, and the actual values are scary. We believe that the only two solutions for making optical discs with a reasonable minimum LE is to either dramatically increase the average LE (to greater than 1,000 years), or to dramatically reduce the standard deviation. Needed is a storage medium that can store data for a minimum of 100 years1.

Conclusions Our research indicates that much improvement is needed for optical discs to have the LE necessary to qualify them as a true archival storage medium. Only one brand seems to have made significant progress in this effort, but there is not enough field data on this brand to include them in the natural aging section of this report.

References 1. 2. 3. 4. 5. 6. 7.

8.

Peterson, Michael, Gary Zasman, Peter Mojica, Jeff Porter (100 Year Archive Task Force), “100 Year Archive Requirements Survey”, January 2007, SNIA’s Data Management Forum, www.snia-dmf.org/100year, p. 1. Lunt, Barry M., Matthew R. Linford, “Towards a True Archival-Quality Optical Disc”, Proceedings of 2009 ISOM (International Symposium on Optical Memory), Nagasaki, Japan, October 4-8, 2009, paper #Tu-G-05. Byers, Fred, “Optical Discs for Archiving”, Information Technology Laboratory, Information Access Division, NIST, OSTA, Dec 6, 2004. Iraci, Joe, “The Relative Stabilities of Optical Disc Formats”, Restaurator: International Journal for the Preservation of Library and Archival Material, 2005, 26:2, 134-150. Navale, Vivek, “Predicting the Life Expectancy of Modern Tape and Optical Media”, RLG DigiNews, Aug 15, 2005, 9:4; also at www.rlg.org/en/page.php?Page_ID=20744#article3 Shahani, Chandru J., Basil Manns, Michele Youket, “Longevity of CD Media: Research at the Library of Congress”, Preservation Research and Testing Division, Washington, DC. Slattery, O., Lu, R., Zheng, J., Byers, F., Tang, X, "Stability Comparison of Recordable Optical Discs- A study of error rates in harsh conditions," Journal of Research of the National Institute of Standards and Technology, 109, 517-524, 2004 . Svrcek, Ivan, “Accelerated Life Cycle Comparison of Millenniata Archival DVD”, Naval Air Warfare Center Weapons Division, China Lake, CA.