Re: Very poor restore performance

ADSM'ers,
     I forwarded this question on to another ADSM team member.  Here is his
response.
Lisa
---------------------- Forwarded by Lisa D. Galindo/Winchester/DST/US on
03/17/99 02:14 PM ---------------------------
03/17/99 02:14 PM ---------------------------


Jay Vandermeer
03/17/99 01:37 PM

To:   Lisa D. Galindo/Winchester/DST/US@DST
cc:
Subject:  Re: Very poor restore performance  (Document link not converted)

They need to generate the statistics to see where the time was spent.

Add to DSM.OPT on the NT machine:
traceflags INSTR_CLIENT_DETAIL
tracefile somefile.txt


The data will be in whatever file tracefile pointed to.

Then run the backup/restore with the commandline interface only - this will not
work through the GUI.




Lisa D. Galindo
03/17/99 11:23 AM

To:   Jay Vandermeer/Winchester/DST/US@DST
cc:
Subject:  Very poor restore performance

Did you have anything for this?

Lisa
---------------------- Forwarded by Lisa D. Galindo/Winchester/DST/US on
03/17/99 11:23 AM ---------------------------
03/17/99 11:23 AM ---------------------------


"Seibt,Stefan HG-Dir ws-sp" <SeibtS AT ALTE-LEIPZIGER DOT DE> on 03/16/99 
07:33:12 AM

Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>

To:   ADSM-L AT VM.MARIST DOT EDU
cc:    (bcc: Lisa D. Galindo/Winchester/DST/US)
Subject:  Very poor restore performance




Hi there,

We have an ADSM server for each operating system client platform, i.e.
an ADSM server for NT clients, an ADSM server for netware clients, an
ADSM server for several unix client. We use a fourth ADSM server only
for archiving files.
Our ADSM server environment describes as follows : OS/390 2.4, TCP/IP
3.2, ADSM 3.1 Lvl 2.1, DB-sizes (NT 14 GB/70% util,Netware 6 GB/40%
util, Unix 4 GB/70% util), BUFPoolsize 16384, LOGPoolsize 4096,
collocation is enabled, our backuppools on DASD (EMC 3390) are migrated
every day to MagStar tapes until the low treshold of 0% is reached,

Before we have upgraded to Server Version 3, the performance of the
backup processes was poor, now the performance of backup is ok and the
recover peformance is still bad. Last week we had to make a disaster
recovery of a NT server. The amount of data was 7 GB and it took 25
(in words: twenty-five) hours. This is about 85 KB/s.

Does anyone have experience with such a behavior and wants to share ?
Any ideas are welcome.

        :-)     Stefan Seibt
                Alte Leipziger Versicherungen
                Abteilung WS-SP
                Alte-Leipziger-Platz 1
                61440 Oberursel

                Tel: +49 (0)6171 66-2375
                Fax: +49 (0)6171 66-7500 2375