As you are seeing, dealing with the metadata for so many files takes an 
excessive amount of time.  The best solution is to forego TSM (or any other 
file level backup product) and use replicating storage, which does block level 
backups.  Snapshots on the storage device will allow you to implement retention 
policy.
 
If you must backup the files to TSM, you should check out image backups.
 
Orville L. Lantto

________________________________

From: ADSM: Dist Stor Manager on behalf of Herrmann, Boris
Sent: Tue 7/3/2007 08:28
To: ADSM-L AT VM.MARIST DOT EDU
Subject: [ADSM-L] Backup/Archive our Document Management System



Hi all,

I'm searching for a good idea how to backup up or archive our important
Document Management System (AIX JFS2). Today we have backed up about 190
Million Files. The Filestructure is the following:

There are Filesystems (40 GB) for every mandant on this system:

<mandant-a>/00001
<mandant-b>/00001
...
<mandant-k>/00001

In every Filesystem we have circa 1 Million Files.


When a  "mandant"-Filesystem ist full the following happens:
-The Filesystem was remouted to the state READONLY
-A new empty Filesystem (40 GB) is created for this mandant with an
increased 5-digit number . New documents are now going to this
filesystem.

What is the best way to backup / archive this system? Some thoughts: Our
backup runs today over 40 hours, TSM DB is growing and growing, and with
backup we've had the problem, that after a crash one Filesystem was
corrupted and after fsck some files/blocks (sorry I'm not a AIX Admin)
are going to the directory "lost+found". But what happens with the
"corrupted" files. I think they will be backed up again and the "good"
backed up Files on TSM expire? If so, this would be a disaster for our
company.

Any hints / suggentions on this issue are appreciate.

Boris