What the Experts Say
Paper and Data Safes coming in 2006 made with: |
 |
|
Long-Term Preservation of Digital MaterialsDr. John W.C. Van Bogart
|
How Long do Digital Storage Materials
Last?
I will now begin the section of the presentation which discusses the longevity of digital storage materials.
At the National Media Laboratory, a variety of magnetic tape and optical disk media types have been aged in accelerated environments. Subsequent failure analysis and modeling of deteriorating media properties allow the longevity of these media to be estimated.
A basic accelerated methodology includes:
- Use of several accelerated temperature/humidity/pollutant environments to artificially age the media
- The measurement of media physical properties and error/dropout rates at regular intervals.
- The modeling of properties versus time using appropriate kinetic models.
Determination of the Life Expectancy, commonly referred to the LE value for a particular medium involves:
- Extrapolation of models determined from accelerated aging studies using standard methods.
- Application of an appropriate set of End-of-Life criteria--consisting of a material property and a value at which failure of the tape system occurs. An example of an end of life criteria is a Block Error Rate of 220 commonly used to evaluate the longevity of CD-ROM media.
Although the life expectancies indicated on subsequent slides are based on experimental results, it is best to use the LEís as "guidelines" rather than numbers "carved in stone." Currently, there are no accepted standards for determining the LE (life expectancy) of magnetic tape media. Standard methods for determining the life expectancy of optical disks have been developed by ANSI, but not all manufacturers perform the necessary product testing.
The prevalence of media vendors, hardware suppliers, and tape formulations make it impossible to come up with a single LE number that is valid for a given media at a specified storage condition. Furthermore, media may fail for reasons other than those considered when determining the indicated LE values. As such, the LE numbers on subsequent charts truly represent an educated guess at this point..
Keep in mind that LE values are similar to "mile per gallon" ratings for automobiles -- "your actual mileage may vary" depending on the number of times the tape is accessed during its life, the quality of the conditions in which it is stored, and the care with which it is handled/shipped.
Modeling the hydrolysis of tape binder and application of an appropriate set of failure criteria allow the LEís for a high grade VHS to be estimated at various storage conditions and temperatures in the chart above. For the sake of discussion, it is assumed that the LE of this particular tape is 10 years when stored at 72 F & 50% RH. Furthermore, only the detrimental effects associated with hydrolysis of the binder are considered in determining LE values.
Cooling tape results in the slowing of the chemical processes which are degrading the tape. Drier conditions reduce the amount of water absorbed by the tape and also reduce the rate and ultimate extent of binder hydrolysis.
One point that this chart makes is that slightly higher humidities can be tolerated at lower storage temperatures for the same level of binder degradation. It also shows the benefits of lower temperature and humidity storage.
The chart also demonstrates that the life expectancy of magnetic tape can be significantly increased by slightly lowering the temperature and humidity of storage. Life expectancies of 10, 17, and 28 years are estimated for storage conditions of 72 F & 50% RH, 68 F and 40% RH, and 65 F & 30% RH, respectively. In these examples, both the humidity and temperature have been reduced simultaneously. However, it is not necessary to reduce both to see a benefit in media longevity. Either the temperature or humidity can be reduced with benefit to the media.
Disposition charts have been developed which can assist archivists in determining the storage conditions necessary for maintaining digital media for a necessary length of time. The chart above considers magnetic tape, optical disk, paper, and film media types. In determining the LE numbers, all of the most likely modes of failure have been considered. For magnetic tape, binder hydrolysis, pigment deterioration, and backing instabilities have all been considered.
Media quality can vary appreciably from vendor to vendor. Also vendors have good manufacturing days and bad ones. Therefore, a range of life expectancies is provided (shown in yellow) to consider these variation in media quality.
On the chart above, areas in green indicate that ALL media should be able to survive reliably under the storage conditions indicated. Areas in yellow indicate that only the "best" vendors will survive for the times indicated. Areas in red indicate that even the "best" media will be unable to survive for the times indicated.
For example, under these storage conditions of 20 C and 40% relative humidity, only the "best" 3480 tape vendors are expected to be able to survive for 20 years. The information for the 3480 tape media appears in the fourth column from the left. A red horizontal line appears on the chart at the 20 year mark for purposes of instruction. The 20 year red line falls on the border between the RED (fail) region and the lower end of the YELLOW region corresponding to the "best" media.
Here is another Life Expectancy chart similar to the one show previously.
However, this time the temperature and humidity on the chart have been reduced.
Continuing with the instructional example regarding the 3480 tape media: Under these storage conditions, there is still a possibility that some 3480 cartridges may fail before reaching 20 years of age. Note that the red line falls through the middle of the yellow section under these conditions. (A red line horizontal line appears on the chart at the 20 year mark.)
This chart shows that the "best" 3480 tapes will survive for an estimated 50 years under these conditions, but the "worst" 3480 tapes could fail after 10 years.
Continuing with the instructional example:
If the storage conditions are reduced to 10 C and 25% relative humidities, the chart now indicates that all 3480 tapes should be able to survive for at least 20 years.
In other words, under these storage conditions, it is estimated that ALL 3480 cartridges in a collection, even the least chemically stable ones, should survive for 20 years.
Several criteria are required to fully qualify a Life Expectancy value.
The same set of experimental data can provide different LE values depending on the end-of-life criteria used. In his book, Optical Recording: A Technical Overview, Alan Marchant states that "one company has variously quoted lifetimes of thirty, fifty, and one hundred years for a write-once optical disk product, basing these different estimates on exactly the same test data." When reviewing manufacturersí media archival life values, it is important to understand the method and criteria by which the LE values were determined.
The "weak link" in a data storage product will determine its lifetime. A tape with a 80 year expected life based on magnetic property performance may have an actual life of only 30 years by reason of poor binder stability.
Digital tape and disk media require special hardware in order to be read. Even if the media survives for 50 years or more, the technology needed to read the data may not survive for more than 10 or 20 years. Thus, the life of a digital archive may be limited by obsolescence of the drive hardware rather than the stability of the media.
You can currently purchase WORM disks which are guaranteed for 100 years. However, considering the explosive growth of CD-ROM and CD-R technologies, it is doubtful that WORM technology will be viable in 10 more years. WORM disks will undoubtedly outlive WORM technology.
Technology WILL change -- Migration to new recording systems and transcription of records to new media is inevitable to preserve the records. Some US government agencies transcribe all of their data to a proven "state-of-the-art" format every ten years. The old records are maintained until the next transcription occurs. In this way, the media/system is only required to last 20 years. Furthermore, there are always two copies of the information available.
Keep in mind that superior technologies do not always succeed. Also, a manufacturer does not have to go out of business for a format to be abandoned. Hardware manufacturers are in business to make money; if a line of equipment is not profitable, or has too small of a market share, it will be discontinued. The above points are true of the obsoleted BetaMax video format.
To ensure that the media will be readable far into the future, it may be necessary to archive system components along with the media. For a 100 year life, recording systems and sufficient spare parts will need to be archived along with the data storage media. Media with life expectancies greater than 20 years are capable of out-surviving existing recording system technologies.
It is inevitable that technologies will advance. Hardware manufacturers need sales in order to grow their business. The more that they can develop and sell technology, the better off their businesses will be.
To help put this into perspective: If I gave you a Word Perfect 1.0 document created in a DOS environment on an 8" floppy disk, would you be able to read it? Do you have the drive hardware, the software, and the operating system necessary to do so?
