Richard,
 
Thanks for the reply.  Here's all I am seeing in the actlog, however during 
this period I am seeing data being transferred from node to the server.  I have 
tried this with no other jobs (backup or otherwise) running with the same 
results; so I don't think its a preemption of this backup job.  I will increase 
the idletimeout to see if it helps.
 
01/18/08   06:59:59      ANR0482W Session 49163 for node EPICREL01DB (AIX)
                          terminated - idle for more than 45 minutes. (SESSION:
                          49163)
 
Both these system are connected to the same Cisco 4000 Gigabit Switch and are 
on the same network [so no routing].  Both machines are EtherChanneled (AIX) to 
a bonded interface on the 4000 switch; essentially 2Gbps network connections.  
I am not seeing any errors on the switch ports or on the AIX system.
 
TIA.
 
Mahesh


>>> Richard Sims <rbs AT BU DOT EDU> 1/18/2008 09:48 >>>
Mahesh -

You need to look in your TSM server Activity Log to see why the
session was disconnected.  You may be constrained on drive resources
such that the very long running, large file backup is pre-empted by a
higher priority process.  Check networking to see if, possibly,
routing might go over a faster path, or if parameters like
TCPwindowsize might be tuned, to reduce the exposure.  Also consider
boosting your IdleTimeOut to 60.

   Richard Sims