I did not get any answer from the list to my original question.
One suggestion, thanks for that, was the I didn't select the
proper device for writing the desired density. But this is not the case.
To conclude the issue I append an answer from the manufacturer:
Answer from an Exabyte specialist:
Looking at your test results it seems that there is indeed an issue
with your Fuji tapes. On the 160m Exabyte tapes you get full capacity.
On the Fuji tapes you get only half of the capacity you might expect.
This seems to indicate that the drive has a hard time writing to these
Fuji tapes.
Is this a fault of the Eliant drive? I don't think so. It performs up
to spec when using the Exatape media it is designed for. As I
explained before, we have no control over the specifications of other
vendor's tapes. It might well be that the Eliant has narrower
tolerance for tape quality than the 8500. It may be that Fuji tapes
can't handle being written to at the high speed used by Eliant (The
drum rotates at twice the speed of a 8500). ...
Exabyte does not do compatibility tests agains non-Exabyte
! tapes. Exabyte *does* design their drives and media to perform
! optimally with each other. If you want to, use our EXPERT software to
make the comparison for yourself.
When I mentioned the tapes getting better over the first few passes, I
meant that rewrites would trend downwards from, say, 2% to 1%. On a
5GB tape this means initially getting (5 - 0.10) = 4.9GB and after a
few passes (5 - 0.05) = 4.95GB. The results from your tests are in
line with this.
Your supplier is right when he claims that Eliant is a better product.
Not only does it support 160m tapes, it also reads/writes at double
the speed of your 8500, and it has a much higher MTBF due to fewer
mechanical parts.
What do I learn from this:
When buying the tapes I did not carefully enough consider all
boundary conditions...
Otto Titze
-------------------(the original question:)----------------------
Subj: Experiences with Exabyte Tapes?
May be the following question is more hardware than Unix related,
sorry about that, but perhaps I can hear about some experiences
We connected 2 Exabyte drives (Eliant-820/EXB 8507) to an AXP 4/166
workstation under Unix V4.0-B. (First we got some wrong information
how to generate the driver. But then I got the proper entries for
ddr.dbase so that I don't believe that it is a software problem any
more)
We have written several 100 tapes with experimental data on other
Unix systems with EXAs 8500. Every tape has about 4.5 to 5 GB of data.
The tapes we used are QG-112M Fujifilm. We had no problems until now.
When we tried to make backup copies of this data tapes on this
workstation on brand new tapes of the same kind, exept in a few cases
the target tapes were full after transfering about 2.5 GB.
(Always cleaning etc.. Reading of our data tapes was no problem on
the Eliants)
Analysis of such a tape by people from Exabyte gave the answer
"The format used is 8500c. as you use a 112m cartridge with a
max capacity of 5gb uncompressed, and as you saved up tpo 6.7gb of
data on the tape, it is quite normal.
I did a transfert of the data on another tape and found out that
the ECC use was twice as much as the maximal admitted."
with the recommendation "use the original Exabyte tapes" (a factor of
two more expensive - and we need several hundreds!)
Indeed tests with original Exabyte tapes gave
- on the 160m Exa-tapes I could write 7.6 GB of data
Regarding 112M Tapes this means that 5.2 GB would fit on a tape -
this would cover our needs
- There was an othe argument that tapes because of dirt during fabrication
are getting better during the first 10 times of usage - therefore
I used one tape several times:
1.
2. 2.467 GB Data copied to it
3. 2.489 GB (cleaning also done)
4. 2.504 GB
5. 2.509 GB
Ok it got better, but less then I hoped ...
My question now is, is this really only a quality problem of the
tapes we used (QG-112M, Fujifilm)
- we have written several hundreds of these tapes on normal EXA
8500 drives (without problems) with up to 5 GB (just these
ones we want to copy now)
- therefore they have proven their quality already
or
- could it be that the firmware of the new Eliant-820 /EXB 8507
is somewhat over-sensible?
(see the above mentioned analysis "the ECC use was twice as much as
the maximal admitted.")
(Unfortunately the new Eliant drives were offered us as a (better)
replacement of the old 8500s, which we could not buy anymore)
Otto
-------------------------------------------------------------------------
| Dr. Otto Titze, Kernphysik TU, Schlossgartenstr. 9, D-64289 Darmstadt |
| titze_at_ikp.tu-darmstadt.de Tel: +49(6151)16-2916,FAX:16-4321|
-------------------------------------------------------------------------
Received on Mon Nov 17 1997 - 19:17:03 NZDT