This library is also single-ended not differential. I am not a SCSI expert
so I am wondering why you would not just leave the library in single-ended
mode instead of changing all of its guts to work with differential. What is
the gain other than the distance you can run your cable? Breece Hill still
offers both models and I was planning on staying single-ended. Although I
am always open to changes for the better:)
>From what I am getting about the new media is that first I need to define a
new device class and specify "drives" for the format to get the maximum
capacity out of the media. Also that data written on filling tapes will
remain in the same format which makes sense. Only data written on new
scratch tapes would benefit from the new drives writing format. That also
make alot of sense.
Assuming that there is no problem reading the media what confuses me a
little is why I would want to define a new device class. In the current
environment the device class is "DLT" and the format is set to "drives."
Those are the only 2 that really matter, correct? As long as I associate
the new drives to that device class I should not have to create any new
device classes. I could just use the existing class and let the drives do
the translations from one format to the next. Am I being too simple?
Jack
-----Original Message-----
From: Paul Zarnowski [mailto:[EMAIL PROTECTED]]
Sent: Wednesday, November 01, 2000 2:16 PM
To: [EMAIL PROTECTED]
Subject: Re: DLT Drive Upgrade
Jack,
We also have a Breece Hill Q47 with 3 DLT4000 drives. It is no longer
our primary robot, but it used to be. It is true that a DLT7000 or DLT8000
drive should be able to read tapes written by a DLT4000 drive. I *think*
it is true that ADSM will read these tapes also, but I do not have
first-hand experience on that. New tapes that are written will be written
according to the format specified in the device class. Check the
documentation on DEFINE DEVCLASS for more info - look for the section on
DLT drives.
However - the main reason I am responding is the following. When we
looked into upgrading our Q47 to DLT7000, we found that financially it did
not make sense. This is because our robot, and drives, were not
differential SCSI. In order to upgrade to DLT7000 drives, we would not
only have had to replace the drives, but we would have had to upgrade the
library itself to be differential SCSI. The cost of this made it
financially not worth doing. Instead, we opted to purchase a new larger
tape library. I would look into whether your library is differential
already or not, and whether you would also run into this issue or
not. Your library may indeed be differential. I think the reason that
ours was not, was because we purchased it before Breece Hill offered a
differential version of the library.
I believe that new data written to an existing tape, will be written in
the format that is already on that tape. I believe that only newly written
tapes will be written at the higher density, and then only if you have the
device class defined properly.
..Paul
--
At 02:02 PM 11/1/2000 -0500, Palmadesso Jack wrote:
> One of my libraries is getting close to maxing out its storage. Its a
>Breece Hill Q47 with 3 DLT4000 drives. I'd like to replace the drives with
>4 DLT8000 drives. So far I've been hearing it should all be backward
>compatible but mostly thats just from the sales people. Some of the people
>around here think a DLT4000 tape should be readable in 7000 or 8000 drive.
>Is this correct?
>
>If it is correct then is it possible to just replace the drives in the
>library redefine them to AIX and ADSM and continue on? In other words will
>ADSM even notice? We are running ADSM 3.1.2 L40 on an RS/6000.
>
>I am thinking it should be fine but I am mainly wondering how new data gets
>written back out to media? Will new data being written to tape need to
>write on a scratch volume or can it just be written on a filling volume?
Or
>is all of this too much of a hassle and I should just buy a new library and
>migrate the data over to it? Is there any sort of procedure defined
>someplace I could use?
>
>Thanks for any input
>Jack
