SUMMARY: Tape Archive Care..Again

From: Matt Marlow (matt@gilligan.UCSD.EDU)
Date: Tue Jul 26 1994 - 22:20:36 CDT

Hi All,

I just received this reply to my earlier posting, and felt I should share this
wisdom in its entirety with all those interested. Thanks Don!

----- Begin Included Message -----

>From deltam!! Tue Jul 26 11:31:03 1994
Date: Tue, 26 Jul 94 11:32:10 PDT
From: (Don Trimmer)
To: mmarlow@UCSD.EDU
Subject: Re: SUMMARY: Tape Archive Care...
Content-Length: 8273


I apologize for this late response, but I didn't notice your original posting.
Anyway, here's my two cents:

First, my qualifications: Delta Microsystems is both a user and a seller of
8mm tape libraries, drives and media. We have over six years experience with
8mm technology. Delta also uses and sells 4mm, 1/4", 1/2", 3480/3490, VHS
and DLT tape technoligies.

The best characterized tape media is 1/2" reel-to-reel tapes. For this media,
studies have found that a tape may be written and then ignored for about two years
before "print-through" begins to cause data loss. If the tape are re-tensioned
once/quarter, then the life of the tape increases to about six years.

I am not aware of any similar studies that have been done for 8mm tape. However,
the 8mm tape drives perform automatic read-after-write when the data is written
to tape. Further, the drives writes about 400 bytes of ECC code for every 1 KB
of data. During the write operation, if any 1 KB data block and it's ECC code
contain any bit errors, the data and ECC code will be rewritten to tape. This
allows the drive to automatically detect and correct several hundred bit errors
in each 1 KB of data during read operations. Therefore, the tape must be seriously
degraded before it can't be read.

We have hundreds of tapes over five years old in our inventory. We have never had
any trouble reading any of them. The datagrade tapes which became available farily
recently have better magnetic properties than the video grade tape we used to use.
Therefore, I would expect it's shelf life to be even better.

I have heard quite a few poeple who claim a much shorter shelf life for 8mm media.
In every case that I am aware of, the real problem was one or a combination of the

  1. Inappropriate media was used. Sony/Exabyte is the best digital media. Media
      from other manufacturers use a different process for binding the oxide layer to
      the mylar substrate. This media may have excellent magnetic properties. However,
      when a piece of the oxide layer falls off, the tape becomes unreadable.

  2. Sony/Exabyte media was used, but the data was written to a marginal tape.
      A few years ago, about 2-5% of new media was marginal. I don't know what
      the current percentage of bad media is, but it's much lower. Marginal media
      can be detected during the write operation by monitoring the "retry count".
      This is available when using our device driver (smt) or through the "mt status"
      command with Sun's "st" device driver. Exabyte says a retry rate of 2-4% is
      acceptable. My personal opinion is that a retry rate of over 1% is not
      acceptable. The retry rate is "(# retries / # KB of data written) x 100".

  3. The drive was dirty our out of alignment when the data was written and/or
      when the data was read. If the drive was out of alignment or dirty (causes
      tape slipage) during the write operation, then the data tracks will not be
      in the correct location on tape. The same problem exists when reading tapes.
      If the drive is dirty, then a tape that was just written may not be readable
      (it would be very fortuitous if the tape slipped the same amount in the same
      places). The problem of an out of alignment drive is a little more subtle.
      A tape that was just written can be read. The problem is that as the drive
      drifts further out of alignment, that tape can no longer be read and it can't
      be read by a drive that is correctly aligned (i.e. it's unreadable).

      Retry rates during write operations and/or error detection and correction
      rates during read operations (also available through "smt" or "st") show a
      marked increase when a drive gets dirty or out of alignment. I'm not ready
      to claim that this is a fool proof detection method, but it seems to work
      very well.

We believe that our customers have written over one million 8mm tapes during the last
5+ years. I don't believe any of our customers are re-tensioning tapes (even though
we used to recommend that they do so). I am only aware of one instance when the
customer avoided the above problems and still encountered an unreadable tape.

Conclusion: While I used to strongly recommend that customers re-tension their
tapes if they planned on keeping them longer than two years, empirical data has
shown this to be un-necessary. I believe that the shelf life (without re-tensioning)
of the video grade 8mm tapes is in excess of 5 years. I belive the more recent
data grade 8mm tapes will exhibit a shelf life (without re-tensioning) in excess
of 10 years.


 | Don Trimmer |
 | e-mail: |
 | voice: 510-449-6881 |
 | FAX: 510-449-6885 |
 | snail-mail: Delta Microsystems, Inc. |
 | 111 Lindbergh Ave. |
 | Livermore, CA 94550 |

> >From uunet!!sun-managers-relay Mon Jul 25 18:42:53 1994
> Sender: uunet!!sun-managers-relay
> Date: Mon, 25 Jul 94 09:40:59 PDT
> From: uunet!gilligan.UCSD.EDU!matt (Matt Marlow)
> Reply-To: uunet!gilligan.UCSD.EDU!matt (Matt Marlow)
> Followup-To: junk
> To: uunet!!sun-managers
> Subject: SUMMARY: Tape Archive Care...
> Content-Length: 2587
> Original Message:
> Hello All,
> It has been suggested in our lab that ALL archive tapes (Exabyte 8mm) be retensioned on a yearly basis in order to prevent damage due to constant magnetic fields. The argument is that, over time, the spooled tape is exposed to a constant magnetic field, caused by its own recorded magnetic pattern, that acts to limit the life of the tape. The proposed solution is to retension the tape - the idea being that it is very unlikely to return to its exact previously spooled position, hence exposing each portion
of the tape to a new magnetic field.
> Has anyone ever heard of this procedure? Is it necessary? Thanks for your help...
> Results:
> Responses were many, but the majority claimed this technique DOES help to
> extend the life of spooled magnetic tapes (1/2" reel. 1/4" cartridge, etc.).
> A few people claimed they had never experienced difficulty in reading their
> archived backup tapes, even though they had been untouched for years. Lucky?
> Quite a few people believed they had lost archived data due to this
> "imprinting" of various magnetic fields, and a few others felt they lost
> data purely because the lifespan of their Exabytes is only a couple of
> years! (I'm not sure what the life expectancy is really supposed to be for
> Exabytes, but I would assume it is at least a few years! After all, what good
> is an archive tape that only lasts 1-2 yrs?)
> It was also mentioned that 'mt -f /dev/rstx retension' does not work on
> systems running 4.1.3 or Solaris 2.3. It is true! But the work-around
> seems to be reading to the end of the recorded data, then rewinding to
> the start of tape as follows:
> mt -f /dev/rstx eom
> mt -f /dev/rstx offline
> Looks like we'd better start "walking" our tapes every now and then!
> There were too many responses to list individuals names, so I would like to
> extend my thanks to all those persons who took the time to respond! Hope
> this helps...
> ////
> (. .)
> *------------------------------o00-(_)-00o--------------------------------*
> Name: Marlow, Matthew Michael Nickname: Matt
> Dept: Neurosciences Phone: 534-5560
> Office: 3047 BSB Postal address: 0608
> E-Mail: mmarlow@UCSD.EDU
> _________/ ___ \ ____ \ __/ __ | __ \
> __/ _/ _/ __/ __/ __/ _/ _| _/ __/
> ________/ ____ < ______/ _____/ __/ ____ | ___ <
> __/ __/ _/ __/ __/ __/ _| _/ __/
> __________/__/ __/__/ _________/__/ __|______/
> *-------------------------------------------------------------------------*
> ----- End Included Message -----

----- End Included Message -----

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:09:06 CDT