Sorry... didn't see the initial email earlier...

As mentioned by Richard, the details on the TSM server and client platforms
& code levels would be helpful.

I seem to remember older clients only initiating a single thread per mount
point on retrieves (but I could be remembering incorrectly).

Also are you retrieving from the primary storage pool or are you retrieving
from a copy storage pool.

It is a single file system but is it many directories and at what level do
the directories branch out?  (that is, are all the files off the root or off
the root is there many subdirectories and are the files spread out across
all of those?)

Within TSM various things that are used as "KEYS" in the TSM data base are
things like the node name, the file system name, the directory path, and
then finally the file name itself.  TSM in trying to minimize lock
contention within the data base will adjust its threads along those
boundaries.  Just something to keep in the back of your mind as you do
things...



**** initial email ****


I archived 1000 files from a single file system using multiple sessions with
resourceutilization 7 and MAXNMP 4 which gave me 4 tape sessions.

When I retrieve this same filesystem with the same parameters I only get 1
session, one tape.    A 2 1/2 hr archive takes 12 hrs to retrieve.

Anyone have an answer?

thanks

Roger

-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Richard Sims
Sent: Thursday, May 21, 2009 6:38 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: [ADSM-L] using dsmc to retrieving large number files

In the absence of details, the List membership can only postulate what
factors may be at play...anything from tape seek delays to network
routing.  You'll have to assess the Archive vs. Retrieve statistics
and server Activity Log entries in conjunction with your knowledge of
the configuration there to gain more perspective on all contributing
factors.  One thing that's certain is that there's more overhead in
creating new files than in reading existing files - worse in some
operating systems than in others.  And, realize that MAXNUMMP is in
play at Archive time, but not Retrieve time.

     Richard Sims