You have requested a page that has been archived. It has been left at this URL to ensure the persistence of links that have been made to it from other resources. It may contain broken links and out-of-date information.
You have requested a page that has been archived. It has been left at this URL to ensure the persistence of links that have been made to it from other resources. It may contain broken links and out-of-date information.
The digital storage media examined are magnetic tapes and optical disks (including magneto-optical disks). For each medium the claims of manufacturers about their longevity, results of accelerated aging tests, and observations from field sites are presented. Recent research, including that presented by Jeff Rothenberg and by the National Media Laboratory, St. Paul, Minnesota, is noted.
The paper concludes that there are at present too many unknowns to commit digital data to currently-available artefacts for anything other than short-term storage. The preferred option is to direct preservation efforts towards solutions which preserve the information content - the digital 'object' - rather than the digital 'artefact'.
The Issues
First, some working definitions:
Preserving the media on which information is electronically recorded is now well understood to be a relatively short-term and partial solution to the general problem of preserving digital information. Even if the media could be physically well-preserved, rapid changes in the means of recording, in the formats for storage, and in the software for use threaten to render the life of information in the digital age as, to borrow a phrase from another arena of discourse on civil society, 'nasty, brutish and short.[3]Archivists have been forced to develop strategies to preserve electronic records because they have had in their care for many years significant quantities of such records. In 1988 this thinking was put into print by the American David Bearman:
We must begin by accepting the information life of specific recording formats as a fact of physics. While we can influence the production of new media and formats and encourage current information recorders to use formats with longer lives, the 'format life' of any given format is the outside boundary beyond which we cannot rationally plan to retain the information without transforming the medium. [4]The Commission on Preservation and Access's Annual Report July 1, 1991-June 30, 1992 summarises the issues for libraries. Its President, Patricia Battin, argues that 'we must remove the burden of archival copy from the . . . artefact' and refocus on 'the concept of managing continuing access to information stored on a variety of media and requiring a variety of ever-changing access hardware and software.' To this end she believes that 'preservation policies must now focus, not on the permanency of the medium, but on the management of permanency in the digital environment.' [5]
So, in relation to electronic records, we are rapidly moving away from the conventional preservation approach - attempting to preserve the artefact. With the rapid rise of electronic records, it is now obvious that the notion of 'saving object X for Y years' will become obsolete, or perhaps applied only to specific categories of information-carrying artefacts such as the book. The primary questions become those of what is worth keeping, and for how long; only when we have answered these can we look at the question of what medium to convert to. The issues here are fragility of each medium, rapid rate of obsolescence of the operating apparatus (software, operating system, etc.), the ease of altering the data, ownership of information (copyright, etc.), and who takes responsibility for its preservation.
As earlier indicated, my role in this paper is to concentrate on the matter of what medium to convert to. To address this question requires knowing about the physical and chemical makeup of the media and about optimum conditions for their storage and handling. Information about these is available: but it is not readily accessible to the consumers (by which I mean librarians, archivists and indeed anyone without the scientific and technical education to understand this information). This point is made by the writers of another report initiated by the Commission on Preservation and Access, this one entitled Research on Magnetic Media-Phase 1:
While there is ongoing research and data available on durable or 'robust' magnetic media, archivists and librarians do not have ready access to this information and are generally unable to interpret the technical data resultant from the research. We know or understand little about the nature of the media, how they react to ambient conditions in storage and during use, and how these properties relate to the long-term preservation and use of information recorded on magnetic media. . . . In order to make decisions regarding migration of recorded information . . . archivists and librarians need to be able to predict the life expectancy of magnetic media when stored under a diverse and variable set of environmental conditions'. [6]In short, the key points are:
Magnetic tape 1 year equipment obsolescence 5 years Videotape 1-2 years equipment obsolescence 5 years Magnetic disk 5-10 years equipment obsolescence 5 years Optical disk 30 years equipment obsolescence 10 years.Rothenberg's estimates caused a flurry of refutation, including a letter from staff of the National Media Laboratory to Scientific American which indicated that 'the physical lifetimes for digital magnetic tape are at least 10 to 20 years.' [9]
Current advertising in the press presents another story. We cannot blame the readers of popular computing magazines or the computer supplements of daily newspapers if they appear to be confused. A completely unscientific sample of current advertising provides the following:
This paper examines primarily the two main digital storage media, magnetic tape and optical disks.
Physical Structure
Some basic knowledge about the structure of magnetic tape allows a better understanding of its preservation problems. The information is stored in the alignment of magnetic particles which are suspended within a polymer binder. This binder adheres the magnetic information carrying layer to a base or substrate; it also provides a smooth surface to ensure that the tape passes smoothly through the tape heads. Other substances are added, for example a lubricant to reduce friction, and a head cleaning agent.
The binder is the most significant factor which determines the longevity of magnetic tapes. It may soften or become embrittled through hydrolysis, a chemical reaction which requires water to be present for it to occur. The polyester linkages in the binder break - the more moisture in the air, the more likely hydrolysis is to occur. Hydrolysis leads to the 'sticky tape' phenomenon where the binder can stick to the recorder heads and lead to clogging of the heads, dropout and other problems. Another problem is lubricant loss, which occurs with the passage of time, even if the tape is unplayed. Less lubricant means increased friction and the 'sticky tape' phenomenon can occur.
The magnetic particles store data in the form of changes in the direction of the magnetism in the particles. These particles (or pigments) vary in their ability to remain magnetically stable (which directly affects the quality of data recorded) according to what they consist of. Although the most stable are iron oxide and cobalt-modified iron oxide, these are not used for high quality tapes where metal particulate (MP) and chromium dioxide (CrO2) pigments are used because of their superior characteristics for recording higher frequencies and allowing higher signal outputs. Van Bogart notes 'there is not much that can be done to prevent the magnetic deterioration that is inherent in the metal particulate and chromium dioxide pigment types' [11] but indicates that storing the tapes in lower temperatures slows the rate of deterioration.
The substrate is most commonly made of polyester film (Mylar, polyethylene terephthalate, or PET). [12] Polyester film is well known to be chemically stable and will easily outlast the binder; rather, its problems arise from mechanical problems such as stresses on the tape caused by fluctuations in temperature and humidity levels in storage areas. These can result in mistracking during playback. Deformation of the substrate can also arise if the tape is not appropriately stressed when it is wound or rewound.
Other factors which affect data loss include whether the recording is helical scan (for example videotapes) or longitudinal scan (for example analog audio tapes) and, of course, the quality and maintenance of the tape recording device itself. [13]
Improving Longevity
It follows from the above description of physical structure of magnetic tape that ways to improve their longevity are based around:
I refer the readers to Van Bogart's report [14] for detailed descriptions of the care and handling required, and continue here with storage conditions. Because binder hydrolysis is the key factor in tape deterioration, and as this depends on the moisture content of the tape, lowering humidity levels means reduced rates of hydrolysis and lower temperatures slow down the rate of hydrolysis. Similarly, the magnetic pigments degrade more slowly at lower temperatures. Reducing the variation of fluctuation of temperature and humidity levels also assist. Storage at high temperatures (indicated by Van Bogart to be higher than 23oC) increases the tightness with which the tape is packed, thereby increasing the distortion of the tape backing and resulting in an increase in permanent dropouts. Storage at relative humidities higher than 70% can also result in increased tape pack stresses as the tape absorbs moisture and expands. Fungal growth may also occur at high temperatures and humidities.
KEY FEATURE ACCESS STORAGE (storage ARCHIVAL STORAGE
for media that allows (storage that
immediate access and preserves the media
playback) for as long as
possible)
Acclimatisation No Yes
required before
playback?
Media life At least 10 years The maximum possible
expectancy for the media type
Temperature Room ambient (15-23oC) As low as 5oC
Maximum variation 4oC Maximum variation
4oC
Humidity Room ambient (25-75% As low as 20% RH
RH) Maximum variation
Maximum variation 20% 10% RH
RH
Life Expectancies
I still have not addressed the question of 'how long'? Taking the key factor of binder hydrolysis (and he is at pains to point out that there are other reasons why tapes can fail, and that his estimate is capable of qualification in many areas) Van Bogart provides this illustration:
For the kinds of ambient room temperatures in Melbourne, say - 25oC and 50% RH, and assuming an unlikely low level of fluctuation - this table suggests an estimated life expectancy of about 10 years. Living in Singapore - 30oC and 80% RH - I must reconcile myself to something more like 1-2 years.
Physical Structure
All optical disks use basically the same structure, the main difference being the way in which the data is recorded. [18] My examples will concentrate on CD-ROM, but the same physical structure and consequent problems apply more generally.
Videodisks are made by a laser which burns minute holes into a glass master, which is then used to make a metal master from which plastic discs are stamped. An acrylic protective coating is applied. Compact discs are similarly produced by a laser which burns pits into a coating on a glass master from which a metal master is produced. This stamps a plastic base or substrate which is next coated with a thin layer of metal, usually aluminium, and is then covered with a protective polymer (or sometimes lacquer) layer.
The metal reflecting layer is considered to be the most susceptible of the factors, largely because the aluminium usually used is more vulnerable to oxidisation than other metals or alloys. Oxidisation leads over time to corrosion which obscures the distinction between pit and surface (that is, between 0 and 1, the way in which digital data is stored), and the data becomes unreadable. Some manufacturers have used other metals or alloys (platinum or gold, for instance) but the manufacturing costs are consequently significantly higher. The polymer base, whose primary function is to support the metal substrate, can itself be permeable to oxygen, thus affording incomplete protection against oxidisation. It can also contain rough spots or other defects which promote localised corrosion. Although optical disk manufacturers use alloys which are more resistant to oxidisation, this can only retard, not halt, deterioration. The protective polymer coating can also fail and again allow oxidation to occur. [19] The bonding materials need further investigation: how are the layers bonded? What is known about the stability of the process? And the ink used for printing onto the disc has been noted as causing oxidation of the metal layer because it caused breakdown of the polymer coating.
Improving Longevity
As for magnetic tapes, ways to improve the longevity of optical disks are based around:
It is generally assumed that optical disks are less vulnerable to damage caused by poor handling than are magnetic tapes, although we are probably all familiar with the temporary unreadability resulting from fingerprints on a CD-ROM. All information storage media need careful and respectful handling, and optical disks are no exception.
A 3M employee makes the comment that the protective coating on a CD is very thin indeed and 'where there's a fault, normally it is that the seal coat doesn't cover everything and something gets in'. [20] It follows then that anything which can minimise the possibility of a fault occurring is worth pursuing. As one example, it is inadvisable to apply adhesive labels to CD-ROMs. Storage at extremes of temperature and humidity can also, clearly, affect the physical structures: for example, as plastic substrates can absorb moisture and oxidisation of the metal layer can occur as a result, then high humidity conditions should be avoided. For the same reasons storage areas in which temperature and humidity fluctuates, resulting in condensation, should be avoided.
Life Expectancies
Peter Adelstein noted in 1993 that while some excellent studies had been carried out on optical disk longevity, there were at that date no national or international specifications. [21] This still appears to be the case today. At one end of the spectrum is an estimate of three to five years for CD-ROMs reported as the view of NARA (National Archives and Records Administration) in 1992. The main problem 'according to Ken Thibodeau of NARA, is that the aluminum substrate on which the data is recorded is vulnerable to oxidation . . . The plastic that protects the substrate is oxygen-permeable, so it provides no protections against the oxidation process.' [22] In late 1994 NARA was still not considering CD-ROMs as an acceptable archival medium:
NARA views CD-ROM as an acceptable transfer medium for permanent records, but has not yet sanctioned it as an archival medium. This means that federal government records that have long-term or permanent value . . . may be transferred to NARA on CD-ROM media but will not be stored permanently on CD-ROM. Once NARA receives such records on CD-ROM, NARA will copy them onto 3480 class magnetic tape cartridges - the only currently acceptable electronic archival medium for permanent storage. [23]Of more general applicability is a report of accelerated aging studies of CDs carried out at 3M, with input from the National Media Laboratory, in 1992. These resulted in 'a 25-year warranty that assures 100 year life-time at room temperature': that is, the lower estimate takes account of 'general storage fluctuation, as long as it's non-condensing', [24] and 100 years is more like the lifetime to be expected from high quality storage conditions.
Studies are still continuing on the life expectancies of optical disks and will certainly need to continue, especially as new kinds become available and enter into common use. An announcement was made recently that the National Media Laboratory is setting up stability studies for CD-R. [25] As yet no useful results are being reported in the non-scientific, library or archival studies literature.
Saffady [26] summarises manufacturers' lifetime estimates for read/write optical disks:
RECORDING TECHNOLOGY LIFETIME ESTIMATE Ablative technology 10-40 years Thermal bubble 10-50 years Dual alloy 100 years Dye-based 15 years Magneto-optical 10-30 years WORM phase change 15 years Rewritable phase change 10 yearsAlthough these estimates are subject to change as more accurate tests are devised and applied, their implication is clear: the lifetime of optical disks of all kinds, and especially CD-Rs, is greater than the technological obsolescence factor of their recording and playback technology.
The primary requirement is that of ensuring that the trainers (using this term to include educators at all levels, from on-the-ground library staff to teachers at tertiary institutions) have ready access to the kind of interpretation of the scientific and technical data which Van Bogart has supplied in his report on magnetic tape. Although trainers have an obligation to ensure that their knowledge and skills base are kept up to date, their task is not always easy, especially if the required data is effectively lost to them because they lack the scientific background to interpret it. While all trainers in the preservation area should have a modicum of scientific knowledge, they often do not: humanities and social science backgrounds prevail amongst librarians and archivists. The Commission on Preservation and Access provides the kind of interpretation which is required, but perhaps there is also a more local role to be played? One envisages the National Preservation Office taking responsibility for a series of papers aimed at trainers, regularly updated and widely disseminated, which summarise current findings and recommendations.
On a more popular level, the misconceptions that we saw in the 'Tape is out. Optical is in' advertisement need to be countered more effectively than is currently the case. If the person in the street is convinced that their Kodak photo CD-ROM will last for ever (whereas their colour film, they know from direct experience, will not) because advertising has consistently and loudly told them that CDs are forever, so, then the more informed but considerably quieter voice of the librarian or archivist will not readily change their mind. This suggests that a different kind of publicity campaign is needed. Perhaps the computer equipment manufacturers, with a significant interest in more sales of their products, can assist with funding to promote the idea that migration of data is the key to this issue?
All of the above ineluctably points to the need to pose (and answer satisfactorily) three questions:
NATIONAL LIBRARY OF AUSTRALIA, Canberra, ACT 2600,
AUSTRALIA, ABN: 28 346 858 075
Telephone + 61 2 6262 1111; Facsimile +61 2
6257 1703; TTY: 1800 026 372