Farren, I was wondering if you could gather a client instrumentation trace
for me to look at to help diagnose the problem here. If you want to do this
and send it to my IBM userid below, I can see if I can give you some
suggestions for performance improvements. To gather the trace, you need to
place the line "testflag instrument:detail" in the dsm.sys stanza you are
using. This will create a file named dsminstr.report.pxxx, where xxx is the
process ID of the job. You need to do a command line backup (don't use the
GUI), and you need to enter quit after the job before the trace is written
out. The trace is not that large (< 1MB). If you have questions, send a
note to my userid ddcanan AT us.ibm DOT com. Thanks.
At 01:03 PM 4/3/2007 -0400, you wrote:

>> On Tue, 3 Apr 2007 16:23:43 +0100, Farren Minns <fminns AT WILEY.CO DOT UK> 
said:


> I am in the process of setting up some Linux TSM Clients at a remote
> site and am wondering if there are any good resources available for
> performance management as so far they are running painful slowly. I
> have tested some ftp'ing of files and these run fine.

> Does anyone have pointers to nay good documentation that could help
> with this?


Especially for long distance work, make sure that all the different
buffers up and down the line are big enough to hold all the in-flight
data.  If your FTP is working well, then someone has adjusted the TCP
Window size, but there are buffers in the TSM server and TSM client
whose capacity needs to be large enough, too.



- Allen S. Rout

Dave Canan
TSM Performance
IBM Advanced Technical Support
ddcanan AT us.ibm DOT com