ADSM-L

Re: 3590 and 3494

1996-03-26 15:01:54
Subject: Re: 3590 and 3494
From: Francis Dequenne <syf AT ECMWF DOT INT>
Date: Tue, 26 Mar 1996 20:01:54 +0000
Tim,

We have ben testing a configuration comparable to what you propose for the last
few months. After an absolutely pathetic start (for one month, the robot was
unable to run correctly for more than a few hours) we have not experienced any
major incident here. the robot works OK, the drive are reasonably reliable.

Do not expect to be able to use the full bandwidth of the 3590 with ADSM. We
see here (and that has been confirmed by other sites) transfer rates going from
less than 1 to -+ 4 MB/s, depending on the size of the file being sent, the
number of simultaneous transactions, the ability of ADSM to multitask, etc. IBM
is aware of this, and we expect some improvements in the future.

Other issues are:
*       ADSM queues tape transaction requests, (i.e. a mount/dismount operation
will not be started before another one is finished. This definitively reduce
robot performances if you have the dual gripper feature installed).
*       ADSM does not recognise a situation where the robot is not responding
anymore to the server (e.g. because the PC that drives the robot has crashed or
because the connection to the robot (in our case, ethernet) is lost.
*       The serpentine nature of the MAGSTAR tape is not recognised by ADSM,
which can lead to severe performance problems for given retrievals patterns.
(e.g. if you try to retrieve several (bits of) files from the same tape)

Again, IBM is aware of these problems, and we hope to receive from them some
fixes/new features in the not to distant future.

Just a hint: If you are going to get this type of configuration installed, make
sure to have an IBM rep on site for the WHOLE customisation and integration of
the subsystem in your ADSM server. There are (were?) many traps or not very
well documented software "features" to be aware off.

On Mar 26,  2:13pm, PITTSON, TIMOTHY wrote:
<=> Subject: 3590 and 3494
<=>We finally got the go ahead to migrate ADSM V2.1 to an RS/6000 (currently
<=>running on MVS).   I've posted many questions for people already running on
<=>an RS/6000 and have received many helpful responses to which I'm most
<=>grateful.  I have yet another question about tape technology.  We've been
<=>going back and forth between 3590 and DLT technologies... we were leaning
<=>towards DLT because it looked as though we were going to have to put remote
<=>ADSM servers at a couple of locations.  However, this is no longer the case
<=>and it looks like we'll be going with one centralized server. We're most
<=>likely going to go with a 3590/3494 solution.  For those of you running with
<=>this type of configuration, are you satisfied with how it's working ??
<=>(reliability, performance, etc.)   I remember seeing some numbers posted
<=>here about 3590 thruput performance not being as good as it should be.  Is
<=>this still the case and, if so, is this going to be addressed by the IBM
<=>folks at some point ??
<=>
<=>Thanks..
<=>Tim Pittson
<=>pittson1 AT bwmail1.hcc DOT com
<=> End of excerpt from PITTSON, TIMOTHY



--
Regards
Regards

+-------------------+----------------------------------+---------------------+
| Francis Dequenne  | Systems Section                  |      /~~\  /~~\     |
| ECMWF             | e-mail: fdequenne AT ecmwf DOT int      |     /    \/    
\    |
| Shinfield Park    | Tel:    (+44 1734) 499361        |   ECMWF             |
| Reading           | Fax:    (+44 1734) 869450        |   ECMWF             |
| Berkshire RG2 9AX | Telex:  (+44 1734) 847908        |     \    /\    /    |
| United Kingdom    |                                  |      \__/  \__/     |
+-------------------+----------------------------------+---------------------+
<Prev in Thread] Current Thread [Next in Thread>
  • 3590 and 3494, PITTSON, TIMOTHY
    • Re: 3590 and 3494, Francis Dequenne <=