6 Oct
2007
6 Oct
'07
6:49 a.m.
First, I don't think that DVD media is subject to "infant mortality" at
all,
which is not to say that a just burned disc can't be defective. Now a
couple of things that look like infant mortality are possible, although the
mechanisms are not the same as what we usually call infant mortality. One
of these is that data written a burner whose laser power is below spec can
"fade" over time, but this is mostly a problem when using "RW" media
(which
should be avoided for archival and backup uses). Also, a media that is
subject to bending can delaminate .... DVDs are a sandwich of two layers of
plastic glued together, and the data is stored on a dye layer coating on the
inside, at the juncture of the two layers. Delamination is fatal to the
data, and normally occurs from the outside in .... Good advice is to only
use 80% to 90% of the media capacity, leaving the outside edge (which is
where the first damage usually occurs) unused for data anyway.
Second, there is no doubt, I don't think, that dual layer media is
considerably less reliable than single layer media (knowing how dual layer
works, it's hard to believe that it works at all). That said, I have not
yet had a problem with any of the dual layer backups that I've made.
The subject of optical media longevity is one of considerable debate. All
studies by the media makers suggest a life of several decades to centuries,
but some skeptics insist on saying less than 10 years. I have optical CD
media that is now 12 years old that I can still read just fine. I think
that some of your practices are excessively conservative (such as only
applying power when the drives are being used), but will do no harm. The
biggest risk, I think, is burning with a marginal burner (low laser power
and/or bad or just dirty optics). Reading the media with a variety of
drives other than the one used for burning is certainly not a bad idea, but
for most users is unacceptably time consuming.
6 Oct
6 Oct
7:47 a.m.
On 10/6/07, Barry Watzman <Watzman at neo.rr.com> wrote:
First, I don't think that DVD media is subject to
"infant mortality" at all,
which is not to say that a just burned disc can't be defective.
I wouldn't call it "infant mortality", but I had a spate of cheap
CD-Rs about 6-7 years ago have numerous manufacturing defects - there
was visible contamination between the top of the polycarbonate blank
and the reflective layer. In one case, I lost a file due to the
reflective layer flaking off around a speck of dust/plastic
shavings/whatever. I have not seen similar visible defects with DVD-R
blanks.
When writing DVDs, I always do a read-verify right after burning, and
for archival data, I tend to create recovery data with programs like
par2, and burn two to three copies. If it's worth my time to archive,
it's worth spending $1 for more than one copy.
As I said, I lost a file once. That was with CD-Rs, and that was the
first one out of many hundreds of discs burned (presuming they
verified at write-time in the first place). I _have_ had a number of
discs that didn't verify, and that could be due to a number of causes,
from defective media to "cosmic rays". I
have not, however, with that
one previously mentioned exception, lost data (yet?)
from disks that
did verify.
The subject of optical media longevity is one of
considerable debate. All
studies by the media makers suggest a life of several decades to centuries,
but some skeptics insist on saying less than 10 years. I have optical CD
media that is now 12 years old that I can still read just fine.
As do I (just last month, I re-read some discs from 1996 with no errors).
I think
that some of your practices are excessively conservative (such as only
applying power when the drives are being used), but will do no harm.
I recall a discussion here about some of the early burners that left
the laser energized even when not burning, leading to appreciable
"dimming" of the laser diode over many years of powered-on idle time.
I don't think leaving an external burner powered off qualifies as
"excessively conservative" in that light, but as you say, it certainly
won't harm the drive, especially if you burn discs rarely.
The
biggest risk, I think, is burning with a marginal burner (low laser power
and/or bad or just dirty optics). Reading the media with a variety of
drives other than the one used for burning is certainly not a bad idea, but
for most users is unacceptably time consuming.
I've only resorted to read verification in a different variety of
drive for data that I have that I would be especially sore to lose
(such as copies of programs I wrote for the PET and C-64 15-20 years
ago, or old digital pictures). For these sorts of data, I do resort
to numerous copies, "just in case".
-ethan
12:16 p.m.
Ethan Dicks wrote:
When writing DVDs, I always do a read-verify right
after burning, and
for archival data, I tend to create recovery data with programs like
par2, and burn two to three copies. If it's worth my time to archive,
it's worth spending $1 for more than one copy.
This is exactly my procedure as well, but I use QuickPar instead of
par2. I also burn two copies, since high-quality DVD-Rs can be had for
less than $0.50 in bulk.
My typical process is to use Winrar with a recovery record (parity) of
1%, then take the resulting files and use Quickpar to generate as much
parity as necessary to fill up the DVD-R. I then burn, ***and I make
sure that the DVD-R also has a copy of winrar and quickpar on the
disc*** (just in case 20 years from now someone is trying to extract my
stuff and doesn't know what a *.par file is).
As I said, I lost a file once. That was with CD-Rs,
and that was the
first one out of many hundreds of discs burned (presuming they
verified at write-time in the first place). I _have_ had a number of
discs that didn't verify, and that could be due to a number of causes,
from defective media to "cosmic rays". I have not, however, with that
one previously mentioned exception, lost data (yet?) from disks that
did verify.
Yes, that's been my experience as well. If they're going to be bad,
they're usually bad right out of the gate (ie. didn't pass Verify).
Every year, I take my very first burnt CDR from 1995 out and try to read
it. It has always read. I keep my CDRs and DVDRs in a cool dry dark
place, though; not everyone can store them in optimal conditions.
--
Jim Leonard (trixter at oldskool.org) http://www.oldskool.org/
Help our electronic games project: http://www.mobygames.com/
Or check out some trippy MindCandy at http://www.mindcandydvd.com/
A child borne of the home computer wars: http://trixter.wordpress.com/
7 Oct
7 Oct
6:02 a.m.
Jim Leonard wrote:
Ethan Dicks wrote:
Jerome Fine replies:
I noticed the replies by Tim Shoppa and Barry Watzman as well as Ethan
Dicks and Jim Leonard. They were all very helpful.
MANY thanks for everyone who replied. It would seem that my precautions
are at the high end rather than the low end. As for copies, the
Fujifilm DVD-R
spindle of 100 was less than $ 30, so the cost is no longer a factor. I
doubt that
I will use up the spindle at less than a dozen blanks a year before dual
layer media
and drives become standard plus single layer media and drives become more
expensive due to lack of sales. And since the time to collect and
prepare the
files for archive (with their MD5 values) is over an hour and the burn
time per DVD
is only 20 minutes, making an extra copy is very reasonable.
My precautions include producing an MD5 value for each Ghost backup file
which I write to the DVD along with the file name and number of bytes.
Since
each backup file is named CddMMMyy.GHO (dd is the day and yy the year),
the date is also included. After I burn the DVD, I read all the files
to an empty
partition and generate an MD5 value for the new copy which I compare against
the original. In about a week, I will again read the Ghost backup files
When writing DVDs, I always do a read-verify
right after burning, and
for archival data, I tend to create recovery data with programs like
par2, and burn two to three copies. If it's worth my time to archive,
it's worth spending $1 for more than one copy.
This is exactly my procedure as well, but I use QuickPar instead of
par2. I also burn two copies, since high-quality DVD-Rs can be had
for less than $0.50 in bulk. from the DVD
using a READ ONLY DVD drive that I
originally purchased 5 years ago and
use only for that purpose. I don't mind that W98SE requires a reboot
since I have
a faulty hard disk device driver which crashes W98SE after about 5 hours
(also
not a problem since I take a break at least every 4 hours) and I am the
only user.
After I read the files on the DVD for the second time, I will delete the
original.
Then, I suspect that it will be useful to read the December backup for
the last
4 years and re-copy then to a DVD (each are just less than a GB) so that I
have an historical archive.
What are Quickpar and Winrar? Please clarify. If they are similar to
MD5, then
I guess I am already doing the same thing. Actually, I would use FC to
compare
the original and the copy, but I know that FC does not catch all of the
differences
with binary files. Plus, after the original file has been deleted, MD5
is the only
reasonable method I have found of checking the archived file.
My typical process is to use Winrar with a recovery record (parity) of
1%, then take the resulting files and use Quickpar to generate as much
parity as necessary to fill up the DVD-R. I then burn, ***and I make
sure that the DVD-R also has a copy of winrar and quickpar on the
disc*** (just in case 20 years from now someone is trying to extract
my stuff and doesn't know what a *.par file is).
It seems that I have missed a valuable file. Thank you.
The MD5.EXE is not on the DVD, although I do have
the GHOST.EXE file that was used to create the backup
files.
Only Barry commented on dual layer DVD blanks. Has anyone
else had any experience? Are they cost effective as compared
with single layer when the extra capacity is considered.
I have been using the Pioneer 105 drive since 2002 when
my son loaned me his to test everything. The following
year my other son and I shared a new drive for a year
(well he had it 80% of the time and I had it 20%) after
which he decided to buy a faster drive for himself. Prior
to that I had the READ ONLY DVD drive which I still have.
Plus I still have a CD burner from 1997 which also still
works since I have all CD / DVD drives powered off when
not in use.
For those individuals who use a CD to boot foreign OSs,
the Pioneer 105 and the device drivers for W98SE read and
make available to Ersatz-11 the first 16 CD sectors (64 blocks
of 512 characters each is the equivalent for a hard drive)
which I have found to be all zeros for a standard CD.
This allowed me to boot RT-11 from the Pioneer 105 drive
using a raw CD media (as opposed to an actual file on either
the CD or a DVD which Ersatz-11 MOUNTs in the normal fashion)
under Ersatz-11 as opposed to any other CD / DVD IDE based
drive which prevented Ersatz-11 from obtaining access to
the first 16 CD sectors where the boot information for RT-11
(and the file structure for the first RT-11 partition) is
located. I was able to use a SCSI CDROM drive, however,
and Ersatz-11 did manage to see the first 16 sectors on the
media which allowed booting of RT-11 in this case.
For anyone who is interested in creating bootable CD media
for PDP-11 hardware or in emulation under Ersatz-11, I can
probably answer many questions. For SIMH, I haven't any
experience with bootable CD media.
Sincerely yours,
Jerome Fine
--
If you attempted to send a reply and the original e-mail
address has been discontinued due a high volume of junk
e-mail, then the semi-permanent e-mail address can be
obtained by replacing the four characters preceding the
'at' with the four digits of the current year.
As I said, I lost a file once. That was with
CD-Rs, and that was the
first one out of many hundreds of discs burned (presuming they
verified at write-time in the first place). I _have_ had a number of
discs that didn't verify, and that could be due to a number of causes,
from defective media to "cosmic rays". I have not, however, with that
one previously mentioned exception, lost data (yet?) from disks that
did verify.
Yes, that's been my experience as well. If they're going to be bad,
they're usually bad right out of the gate (ie. didn't pass Verify).
Every year, I take my very first burnt CDR from 1995 out and try to
read it. It has always read. I keep my CDRs and DVDRs in a cool dry
dark place, though; not everyone can store them in optimal conditions.
6:21 a.m.
On 10/7/07, Jerome H. Fine <jhfinedp3k at compsys.to> wrote:
Jim Leonard
wrote:
> Ethan Dicks wrote:
>
>> When writing DVDs, I always do a read-verify right after burning, and
>> for archival data, I tend to create recovery data with programs like
>> par2, and burn two to three copies. If it's worth my time to archive,
>> it's worth spending $1 for more than one copy.
>
> This is exactly my procedure as well, but I use QuickPar instead of
> par2. I also burn two copies, since high-quality DVD-Rs can be had
> for less than $0.50 in bulk. My precautions include producing an MD5 value for each
Ghost backup file...
What are Quickpar and Winrar? Please clarify. If
they are similar to
MD5, then I guess I am already doing the same thing....
MD5 generates a simple checksum. Winrar (rar/unrar) splits a large
file into uniform segments with optional data integrity record.
Quickpar (and par2) generate additional recovery records that can be
used to reconstruct a damaged archive. What one might do is to start
with a disk image file, of a, say, 500MB filesystem, then hand that
image file to Winrar (rar), telling it to make as many 20MB parts as
required, then run Quickpar (par2) on the collection of 20MB rar
parts, and to generate, say, 20% excess data, resulting in ~500MB of
rar parts, and ~100MB of par parts, then burn that whole mess to a
CD-R. To reconstitute, one can copy the pars and rars to a DOS or
UNIX machine (or other), run Quickpar (par2) to verify that all the
rar parts are intact, then use rar to rebuild the original 500MB file
from all the 20MB parts. The advantage comes in when
you have one or
more of the 20MB rar parts that can't be read from the CD-R...
you
have previously provided for excess data, so as long as you haven't
lost more than about 100MB of your original data, the par files can be
used with the intact rar parts to reconstitute the missing/damaged rar
parts. One analogy is that this is like RAID 5 for files... as long
as you have a sufficient percentage of the original data and recovery
data intact, you can restore what you are missing.
This allows you to archive your data in a way that allows for less
than perfect retention over time. As long as you catch media
degradation before it passes your self-selected threshold, you can
still recover what you originally wanted to save and migrate it to new
media. For the truely paranoid, you are allowed to save arbitrary
amounts of recovery data, to 100% or more (i.e. - a 500MB filesystem
along with another 500MB or more of recovery data). One could even
save that 100% or 150% of recovery data *without* the original data,
and simply recover the original files when required, but that's an
extreme use of the tools.
I only occasionally go to these lengths for data archiving - for
photographs or software that was never available commercially (i.e.,
my own), since these data aren't likely to be in the hands of anyone
else (unlike commercial software packages and operating system distros
for various platforms).
In short - it's belt-and-suspenders tools for archiving. Not
essential, but the first time you have to restore a missing rar bit
from par files, it makes all the hassle worthwhile.
-ethan
6 Oct
6 Oct
8:10 a.m.
"Barry Watzman" <Watzman at neo.rr.com> wrote:
The subject of optical media longevity is one of
considerable debate. All
studies by the media makers suggest a life of several decades to centuries,
but some skeptics insist on saying less than 10 years. I have optical CD
media that is now 12 years old that I can still read just fine.
While I like to pontificate on the subject too, in my experience
scratches that damage the reflective layer are clearly (no pun intended)
responsible for 95% of CD failures I've witnessed. I too have CD-R
disks in the 10-12-14 year old range that read just fine. I have torture
tested by exposure to direct sunlight at elevated temperatures for
several years at a time and it still reads just fine (I know, not the
same as a real BER test, ...)
I think
that some of your practices are excessively conservative (such as only
applying power when the drives are being used), but will do no harm.
I have found that CD-R burners rarely last more than 3 years. Actually
the latest one I have is on its fourth year... maybe they fixed something
in the past couple of years, or maybe I'm just lucky! (Probably just
lucky because other ones in the farms at work die after 2 or 3 years.)
Many OS's are not happy about IDE or SCSI drives connecting/disconnecting
without a reboot, although I know some tricks for Linux.
The
biggest risk, I think, is burning with a marginal burner (low laser power
and/or bad or just dirty optics). Reading the media with a variety of
drives other than the one used for burning is certainly not a bad idea, but
for most users is unacceptably time consuming.
I think it's wise to do at least spot-checks.
Tim.
6287
days inactive
6288
days old
6 comments
6 participants
participants (6)
-
bfranchuk＠jetnet.ab.ca -
ethan.dicks＠gmail.com -
jhfinedp3k＠compsys.to -
shoppa_classiccmp＠trailing-edge.com -
trixter＠oldskool.org -
Watzman＠neo.rr.com