Same here.  The files are 4K average in size (geological mapping data -
lots of png files).

Have you tried MEMORYEFFICIENT?

Windows 2003 32-bit

Yes, the backup start takes many hours due to getting the meta-data from
the TSM server on what files are backed up.  This can only be done with
MEMORYEFFICIENT on this many files or journaling so the client doesn't
have to download the meta-data.  You have to configure the journal
properly.



From:
Mehdi Salehi <ezzobad AT GOOGLEMAIL DOT COM>
To:
ADSM-L AT VM.MARIST DOT EDU
Date:
06/26/2010 05:42 AM
Subject:
Re: [ADSM-L] incremental backup of many millions of very small files
Sent by:
"ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>



Zoltan,
Actually the node is a Win 2008 32-bit. I said journaling is not feasible
because b/a client incremental backup is not feasible as well. When I
start
backup, nothing happens even after two hours. I don't know how could you
manage to backup 90 millions of file in a filesystem. Perhaps file size
matters, as my files are about 10KB in average.