To respond to your question about archival storage of storage pool
tapes:
It is unwise to put a data tape away for years and never read it.
It can happen with *some* tape that the long-term proximity of the
winds can cause "print through" as the magnetization on one wind
can induce a ghost image copy of that pattern on the adjacent,
contacting wind. (It is a historic problem with audio tape.)
A tape never used is a tape whose condition is an unknown:
Unless you rotationally use your tapes (as ADSM tends to), you just
don't know how usable a tape is. Also, a tape stored away for years
tends to become a tape forgotten. There are countless stories in
data processing of vital, archival tapes getting lost or deteriorated
over time due to moisture, mold, mishandling, etc.
Things change while the tape is frozen in time: While the tape is
stored away somewhere, the software that originally created it is
going through PTF, release, and version changes. The hardware that
wrote it is going through microcode and drive tolerance changes.
Such factors contribute to various degrees of incompatibility which
increase over time. There is good probability that hardware five
years from now may read today's tape, but I would not count on that
future software being able to fully interpret it.
Commercial media is intended to be transient, not archival:
Common technology is not designed to last for long periods - not to
cheat the customer, but simply because of economics and that the
advances in technology cause us to move on anyway. Tapes are almost
totally plastic, which inherently gets brittle over time as its
polymers dry out, which also causes it to change dimensionally, which
thus stresses the oxide coating. By rotationally using and thus
copying the digital data on tapes we both assure that it's readable
and mitigate the aging effects on the minute magnetic domains on the
tape by "renewing" them in regeneration on another surface.
So, your best means of assuring the quality of your media and data
thereon is to keep it circulating.
Richard Sims, Boston University OIT
|