As for the AIX perspective, we too converted everything TSM related, to
RAW versus JFS (we went through the IBM Technical Exchange seminars,
especially the ones that covered PERFORMANCE).

>From what we can tell, things seem to have become worse. We have not seen
any drastic (or just really noticable) performance benefits from RAW.





"French, Michael" <Michael.French AT SAVVIS DOT NET>
Sent by: "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>
01/22/2004 10:12 AM
Please respond to "ADSM: Dist Stor Manager"


        To:     ADSM-L AT VM.MARIST DOT EDU
        cc:
        Subject:        Re: RAW vs. JFS question


        I can't speak to AIX, I use Solaris but I just converted all of
my volumes from mounted, VXFS ones to RAW and the performance difference
has been huge.  I converted one of my servers this past Saturday, I have
10 DB vols and 10 mirrors plus 1 large volume and a mirror.  I started
expiration processing on Sunday and it has processed almost 14 million
objects since then.  It would have taken 6 months or more of doing this
steadily to get the same amount completed.  Also, I had many client
sessions that weren't completing each day do to having really large file
spaces (Solaris clients), they now complete normally each morning
instead of running all day.

        I am using Solaris 8 on E4500 servers with 4 processors and 4 GB
RAM with TSM 5.1.8.1.  I completed converting all of my servers this
past Tuesday and I have seen the same performance gains across the
board.

Michael French
Savvis Communications
IDS01 Santa Clara, CA
(408)450-7812 -- desk
(408)239-9913 -- mobile



-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Gerhard Rentschler
Sent: Thursday, January 22, 2004 9:05 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: RAW vs. JFS question


I think the most interesting question is still unanswered: how much
performance do I gain with raw volumes? More exactly: how much less time
will an expiration take on a 100 GB TSM data base? I don't think raw
volumes would make sense for disk caches.

Does anyone have experience in this area?

Best regards
Gerhard

---
Gerhard Rentschler            email:g.rentschler AT rus.uni-stuttgart DOT de
Regional Computing Center     tel.   ++49/711/685 5806
University of Stuttgart       fax:   ++49/711/682357
Allmandring 30a
D 70550
Stuttgart
Germany