news     faq     contact us     prov

4.3 Media refreshing

When the challenge of preserving digital objects comes up, the first issue that most people identify is the lifespan - or lack thereof - of the media on which the digital object is stored. There are really two related issues here:

• Media deterioration. The physical media on which the electronic records are stored will deteriorate over time and eventually become unreadable.
• Media obsolescence. Even if the media remains readable, it will eventually become impossible to obtain suitable readers to read the media.

The physical media on which the electronic record is stored will deteriorate over time and eventually become unreadable. Part of the deterioration is due to wear and damage as the media is used, and part of the deterioration is due to ageing of the media. Wear and damage occurs to all media. For example, tape wears as it passes around capstans and read/write heads. Even CDs, which do not suffer wear in the same fashion, suffer damage that causes errors due to physical handling of the CD. There have been many examples of slow chemical changes in media causing problems with longevity. Examples include:

• paper-based records on paper with a high level of acid, which would eventually cause the paper to disintegrate
• magnetic tape in which the binder that holds the magnetic particles to the tape decompose and cause the layers of tape to stick together in the reel
• CDs in which poor manufacturing processes allow the aluminium reflective layer to oxidise.

It is possible to reduce the rate at which media deterioration occurs. Wear and damage can be reduced by careful handling and adjusted readers. Chemical deterioration can be reduced by holding the media in a controlled environment. However, it is not possible to prevent media deterioration. Sooner or later, every piece of media will become unreadable.

Short as the lifespan of a piece of media is, the lifespan of the hardware necessary to read the media may be even shorter. A CD may physically last one hundred years, but it will almost certainly not be possible to obtain a CD reader at that time that will read that CD.

Being physical devices, the readers are also subject to mechanical wear and chemical deterioration. Consider a CD reader, for example. The servo mechanism which controls the position of the laser suffers wear and will eventually be incapable of controlling the position sufficiently precisely to read the CD. The laser that reads the CD deteriorates over time and loses power and will eventually be incapable of reading the CD. Deterioration of the readers will be a more significant problem with modern technology than older technology because of the miniaturisation and complexity of modern technology. For example, it is quite feasible to machine a part for a half-inch reel-to-reel tape drive. It will not be economically feasible to grow and fabricate a semiconductor laser of the precise frequency required to read CDs.

The wear and deterioration of media readers means that they will eventually need to be replaced. Unfortunately, hardware has an economic lifespan, and once an item is no longer manufactured the cost of replacement becomes prohibitive. The lack of spare parts usually results in the cost of repair also being prohibitive. Even if it is possible to obtain, or maintain, the hardware, it is usually difficult and expensive to physically connect it to modern computers and impossible to obtain the necessary software to drive it. The economic life of a media technology varies significantly. Widely-used media built by many vendors (such as 3.25 inch floppies) have a long lifespan. On the other hand, expensive, proprietary, media technology can have effective lifespans of only a few years. Modern technology is likely to have a shorter economic lifespan than older technology due its greater complexity and miniaturisation.

The only currently practical solution to the limited lifespan of media and the hardware that reads them is periodic replacement of the media. This involves copying the digital information from one piece of media to another piece of media. This practice is referred to as 'refreshing'. The copy may be to the same type of technology (e.g. the replacement of one hard disk driver with another), or it may involve a transfer to a new technology. Both approaches protect against mechanical or chemical deterioration of the media, while the second also protects against technical obsolescence of the media.

The good news is that media refreshing is a completely solved problem. All IT departments routinely refresh media, and tools and systems to manage and perform refreshing are commercially available to support sites with large amounts of data.

The cost and risk of refreshing is largely dependent on the degree of automation of the process. PROV strongly recommends highly automated systems for managing media for this reason.

The most expensive option is where the media is stored offline on shelves. Refreshing consequently involves manual handling of the media; fetching the original media from storage, loading it into the system, and storing the new media back onto the shelves. It is necessary to take great care in the external labelling and handling of the replacement media, as it is easy to misfile or mislabel media and hence lose records.

Fortunately, media management systems have largely eliminated the need for manual refreshing. It is possible to purchase 'tape libraries' or 'silos' that replace the manual transfer of media from storage to the readers by robots. The great benefit of these systems is the reduced cost of loading and storing media and the reduced risk of misplacing media.

The first risk of refreshing is that the copy will not be exact; that somewhere in the refreshing process the bit stream will be corrupted. Fortunately, it is easy to perform a bitwise comparison of the data from the original piece of media and the copied data to determine the accuracy of the refresh. This, however, will approximately double the time necessary to perform the refresh, and hence halve throughput.

The greater risk of refreshing is the risk of overlooking a piece of media and hence not refreshing it. This raises the risk that the piece of media will be overlooked for so long that it is no longer possible to read the media because of deterioration of the media, or lack of a suitable reader. This risk is minimised by avoiding manual handling, with the attendant risks of misplacing or mislaying media. Media management systems automatically track the location of media. Some types of tapes record the number of times the tape has been used in a chip in the tape cartridge itself.

back to top

printer friendly

VERS Story | Standard | Assessment | Projects | Digital Archive | Training | Toolkit | Publications

Site last reviewed: 20 Jul 2006
© State of Victoria 2008 | Disclaimer | Privacy Statement