I on the other hand disagree!.

The ADSM database does a good job of many of the things it needs to do, but as
always, I'm sure that it is being used in ways for which it was never  designed.
A good general purpose database will be able to cope with such changes better
than a purpose built one. Also, a general purpose database has appropriate
utilities to do many of the things that we are trying to do with our black box.
As an example, think how easy it would be to change platform types with a useful
unload/reload utility.  What about unreasonable limitations on database size
that exist with the black box?

The problem as I see it is one that besets the whole of ADSM. There are some
users  (like me) who back up two servers of one kind and don't want unnecessary
complexity, and there are others who back up 10000 nodes across 50 servers and
wouldn't mind the complexity if it added useful function.
My thinking is that a cut down set-and-forget general purpose database
implementation would suit the first crowd, i.e. the black box of today but with
a prepacked/configured DB2 or Oracle instance, whilst those with somewhat bigger
needs could pay for a TSM consultancy to tune the many knobs on their database,
and the really big guys buy a full version of the same database and go all out
with it.

PS Just so's you know, I'm a DB2 bigot and having now worked with Oracle for a
while would only use Oracle for non-technical reasons such as availablility of
local expertise were I given a choice of database by my application.

Steve Harris
AIX/ADSM/Oracle/HACMP guy
The Wesley Hospital
Brisbane Australia.









"Kelly J. Lipp" <lipp AT STORSOL DOT COM> on 17/11/99 07:21:10

Please respond to lipp AT storsol DOT com








 To:      ADSM-L AT VM.MARIST DOT EDU

 cc:      (bcc: Steve Harris/Wesley)



 Subject: Re: Using db2 or oracle or whatever for the TSM DB



Fax to:




I'm in violent sympathy with Wanda's position.  It is not broken, why break
it!

A general purpose database will not be optimized for this environment.  I
can't imagine that an Oracle could be made to work better than what we have
today.  Can what we have today be made to work better?  Perhaps, but I doubt
that many of us on this side of the fence can say that with any certainty.

Why does anyone think we need to change databases?  Size, performance,
maintainability? What is it?  How many of us really want to administer
another db?

Kelly J. Lipp
Storage Solutions Specialists, Inc.
PO Box 51313
Colorado Springs CO 80949
719-531-5926
www.storsol.com
lipp AT storsol DOT com