28 Jul
2004
28 Jul
'04
6:51 a.m.
----- Original Message -----
From: "Pete Turnbull" <pete(a)dunnington.u-net.com>
To: "General Discussion: On-Topic and Off-Topic Posts"
<cctalk(a)classiccmp.org>
Sent: Wednesday, July 28, 2004 2:22 PM
Subject: Re: Article on data rot on CD's
On Jul 28, 12:11, Joe R. wrote:
Did the table show any data for what the actual failure mode of the CDs was?
Heat and humidity oxidizing the reflective coating will kill a cd very
quickly, but this assumes moisture can get through the protective coating. I
would also like to know what affect printable adhesive labels have on media
along with permanent markers used for labeling. Having a tested 25yr life on
media doesn't mean much if a felt tipped sharpie marker can ruin a cd in a
year or if the glue on a label can do the same. I have cds burned back in
the mid 90's from a HP4020i 2x burner that was $1200 at the time and only
had 1 failure that I know about and that was a backup OS cd I made that got
beat around and scratched for years.
The problem is that AFIK no one has found ANY
CD disks that are
reliable. Several people that have been interviewed in national
publications explictly pointed out that they bought top quality disks
but
they were still unreliable. In fact, it
didn't appear that there was
much
difference between the cheap ones and the
expensive ones.
The other day I came across a table from a report showing the relative
longevity of data on various media (DLT, CD-R, etc) at a variety of
temperatures and humidities. I'll try and find it again and post some
of the results. Some of you might be shocked. For example, a CD-R
with an expected lifetime of something like 25 years (if I'm not
misremembering the highest figure) under ideal conditions has a
lifetime of only several *months* at higher temperatures (upper 20s C,
that would be 80s F) and humidity. DLTs fared much much better.
I have some CDs that were bought about 8 years ago because they were
supposedly good quality, and burned in a highly-rated burner. Out of
the first batch of ten, 4 are now unreadable or give multiple errors.
--
Pete Peter Turnbull
Network Manager
University of York