ADSM-L

Re: Does the Windows journal engine affect (help) ARCHIVE performance ?

2003-06-07 18:19:14
Subject: Re: Does the Windows journal engine affect (help) ARCHIVE performance ?
From: Richard Sims <rbs AT BU DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Sat, 7 Jun 2003 18:18:52 -0400
>I have a Windows2000 client who has well over 5 million files that we
>ARCHIVE every week for offsite-vaulting purposes. ...

Rick - Your situation is, as Mark suggests, trying to accommodate something
       which at least sounds ridiculous, as if a non-technical user defined
the method being forced on the data processing staff.  Doing a weekly full
backup is just sooo 1970s, and contrary to what TSM is all about, in trying
to improve on that ancient methodology.  I think we'd be very surprised if
five million files actually changed every week to require the totality to
have to be shuffled off every week.

>... At any rate, due to the large # of files (currently
>approx 5 million) and the large size of the backup (approx 175GB), the
>ARCHIVE job takes a couple of days to complete.

That statement defines a principal problem that needs to be addressed before
approaching anything else: it should by no means take days to back up 175 GB.
What is vague in your posting is: just what is that "ARCHIVE job"?
It sounds like it's not really the "weekly full" you later mention, but
rather sounds like some oddball hack created for that PC to find and then
send data to the TSM server, which may be a great big performance bottleneck.
That needs to get fixed, first.

You say that the proper copy storage pool approach is untenable for some
reason.  I gather that the data is sensitive and that they aren't doing
incrementals, but rather just creating a primary pool tape set which is
then whisked out of the building.  That's not necessarily bad unto itself;
but absent from your posting is perspective on the ongoing file repertoire
as recorded in the TSM server database: is this a policy set which keeps
only one version, or a bunch that inflate the database?  And we don't know
how tapes come back and their cataloged contents are then treated relative
to TSM.  Your posting also talks of an onsite copy of that data, which
confuses the big picture here.

Your posting is seeking advice on making a technical choice where the
underlying, fundamental requirements are undefined.  A systems analyst
would advise you to start from scratch and fully define the requirements,
and only then consider solutions.  Don't build a house on sand.

  Richard Sims, BU