It seems that there aren't any hard and fast rules on how long to keep an
Exabyte tape around. My original message was:
> I have an Exabyte drive and I am using Sony QG-112M tapes. They seem to
> work just fine. Every week, I put a tape in my Exabyte drive, erase it, and
> then allow nightly incremental dumps onto the tape. I have 6 tapes that I
> am cycling through. I haven't gotten any read-write errors or any other
> problems on the tapes. It has only been 8 months since I started doing the
> cycling. How long can I expect to use tapes in this fashion? What criteria
> should I use to determine when I should replace this set of tapes with a new
> set? Are the criteria real ones or mythology (like all of Anatomy)?
I haven't been able to afford buying a lot of new tapes because I don't have
a "major" grant yet and have been working on a budget where buying a
shoestring can be considered extravagant. ;-( Oh well!
(PS: After 8 months use, I have all 0's on retries on 6 weekly tapes and 5
biweekly systems dump tapes.)
Thanks to all who replied!
From: uunet!deltam!flyer!mark (mark galbraith)
We recommend that tapes be used for 50 passes (read or write) before retirement.
Alternatively, you could use a program such as our smtmon that actually queries
the tape quality at the time of the backup. The results are displayed in a graph
form for you to see the tape quality at a glance.
--Mark Galbraith Voice: +1 415 449 6881--
--Software Engineer UUCP: uunet!deltam!mark--
--System Administrator/Postmaster Domain: email@example.com--
--Delta Microsystems, Inc. Compuserve: 76234,3126--
From: uunet!millidc!indigo!djm (Drew Montag)
I've been cycling the same 4 tapes around for our Monday - Thursday level 9
dumps since February 1989. The number of retries has stayed very constant,
with each individual tape having it's own ratio of errors/MB.
One of the original articles about the Exabyte, by Goble at Purdue, mentioned
an episode where a tape was accidentally left in a drive with the spinning
head in contact for an ungodly amount of time, with no apparent damage to the
tape or the head. I take this to mean that a normal tape should outlast the
computer it's backing up, and maybe the Exabyte drive.
From: uunet!mundo.eco.utexas.edu!tyen (Anthony Yen)
What we do here at UT/Austin is record the total number of blocks
written (dump reports this after it finishes a partition), and
the number of retry errors (an `mt -f /dev/nrst0 status' reports
this). When the retry/blocks ratio exceeds 4%, we clean. When
it still exceeds 4% after cleaning, we ditch the tape. Been run-
ning for six months now on their standard-grade video tapes (you
are using data grade, which is much more durable), no problems.
From: uunet!meaddata.com!keves (Brian Keves - Consultant)
Exabyte tapes do have a practical limit for reuse. What that limit is is up
to you and how you feel about your backups taking longer while getting less
data on your tapes.
Exabytes automatically do error correction so you don't have to worry
about losing data, but each time you use the tape you will get more
errors on the tape. Depending on the quality of the tape and what it
was originally designed for, this error rate could increase dramatically.
Tapes designed for Camcorder use are not good enough for lots of reuses,
but tapes designed specifically for data storage do better.
I have found that Sony tapes are the best for this sort of thing. If you
bought a lower quality tape then you can expect that you will get more
and more tape errors when you reuse them.
The problem with tape errors is that they take time to resolve and use
up tape space. This happens automatically, true, but it can be annoying.
I have seen Maxell tapes lose 100s of MB of space after just 25 reuses.
Exabyte media is soooooo cheap that most people don't bother to reuse the
tapes. Personally I find that this is the best solution. In a large
environment I mixed full and incremental backups onto the same tape and
then just store it indefinitely. This way you can just buy any kind of
tape you like.
From: David Fetrow <uunet!orac.biostat.washington.edu!fetrow>
This is an interesting question. Here we've been buying new tapes after
less than 10 read/write cycles (at $8/tape, why not?). I'd be interested
in how long we could extend that to.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:20 CDT