ADSM-L

Re: Slow Stgp Backup

1998-04-03 11:07:56
Subject: Re: Slow Stgp Backup
From: Hilton Tina <HiltonT AT TCE DOT COM>
Date: Fri, 3 Apr 1998 11:07:56 -0500
Also check these options in your dsmserv.opt file.  Ours were set very
low (I don't know why) and by increasing them to the default the
tape-to-tape copies (both backup & move data) got faster.

        TXNGroupmax, MOVEBatchsize, MOVESizethreshold

Tina Hilton
Thomson Consumer Electronics

        ----------
        From:  Bill Colwell[SMTP:bcolwell AT DRAPER DOT COM]
        Sent:  Friday, April 03, 1998 10:39 AM
        To:  ADSM-L AT VM.MARIST DOT EDU
        Subject:  Re: Slow Stgp Backup

        In
<Pine.OSF.3.96.980403084412.9865A-100000 AT curly.cc.utexas DOT edu>, on
        04/03/98
           at 08:59 AM, Robert Jarry <jarry AT UTS.CC.UTEXAS DOT EDU> said:

        >I am running a Backup Storage Pool for the first time and boy
is that puppy
        >slow!  It has been running for 17 hours and it is still on the
first DLT.
        >The primary pool is on a DLT 4700 and the copy pool is a single
DLT 4000.
        >Both devices are on the same scsi bus but I can't believe that
would cause
        >it to run this slow. Only 49 GB, 189K files, have been backed
up in 17
        >hours. There have been no errors, unreadable bytes, or
unreadable files.
        >The server is v2r1m12 running on AIX 4.1.5 on a R$0 (load avg
4.0).

        >Please, does anyone know how I can speed things up.

        >Thanks,
        >Robert


>-----------------------------------------------------------------------
-------
        >Robert Jarry                             email:
        >Robert Jarry                             email:
r.jarry AT cc.utexas DOT edu Unix
        >Services                            phone:  (512) 475-9330
Academic
        >Computing & IT Services
        >The University of Texas at Austin

        Do a 'q db f=d' and check the cache hit ratio.  If it is less
than 98 add
        more database buffers; in the server options file increase the
dbbufpoolsize
        value which is in Kbytes, i.e. 8192 = 8192k = 8 meg. This will
help overall
        server performance.

        If your cache hit ration is already over 98, then maybe the
problem is in
        your i/o setup.  I can't say for sure since my platform is
os/390.

        You can kill the backup process and restart the server. The
        backup will pick up where it left off.
        --
        -----------------------------------------------------------
        Bill Colwell
        C. S. Draper Lab
        Cambridge, Ma.
        bcolwell AT draper DOT com
        -----------------------------------------------------------

<Prev in Thread] Current Thread [Next in Thread>