Lynn,

        We are a HP shop with 24 large HP-UX servers and 130 NT servers (90%
HP) backing up 7 TB via TSM.  We have been running the TSM server on HP-UX
with ATL tape libraries for almost three years now !  We had one major
problem with an upgrade between TSM, ATL and HP...  It turned out to be a
nasty problem with the HP drivers for DTL7000s, while HP technical support
was pathetic, ATL and IBM were great!

> -----Original Message-----
> From: Sattler, Lynn [SMTP:Lynn.Sattler AT DANA DOT COM]
> Sent: Tuesday, September 19, 2000 10:11 AM
> To:   ADSM-L AT VM.MARIST DOT EDU
> Subject:      Would you buy TSM again.
>
> We are reconsidering our backup software.
>
> We are a pretty large unix shop (5 Terrabytes to backup nightly) and are
> using some HP Omniback and a lot of Veritas Netbackup.  Mostly HP-UX.
>
> We have a large Lotus Notes Rel 5 implementation on AIX (600 gig on its
> way to 1.2 Terrabytes).
>
> We are studying robot technology and SAN's.
>
> Tivoli is selling and Netbackup is trying to stay in the shop.
>
> How good do you rate TSM on a scale of 1 to 10.  (10 being best).
        [Kovacs, Mark]  [Kovacs, Mark]  >9.... nothing is a ten, but I've
used most all of them at one time or another in the past.
> Would you definately buy it again, or would you shop around.
        [Kovacs, Mark]  >Buy again.  We had a recent issue with a very slow
NT restore of 100+ GB and are looking into some fixes to the overall
configuration.
> My gut feeling is that everybody has some complaints with their backup
> solution.
>
        [Kovacs, Mark]  > No major complaints, but here are some
observations:

        1. The person administrating TSM has to be trained on TSM.  There is
a company out in Colorado Springs that does a fantastic job (much better
than you know who....)

        2. You must really spend some upfront time and plan for the entire
implementation.  Really look at your data and your servers and really
understand your nightly incrementals.  I would recommend bringing in IBM
Global Services (their TSM demi-gods in particular) in for the design and
initial setup.  We went initially with some IBM Business Partners and latter
had others in for a system health/reality check.  My Jr. UNIX Admin knew ten
times as much about TSM as their "expert".  TSM is not a canned super TAR or
FBACKUP, but a much more complex and capable solution.

        3. Require (sometimes at gun point) your UX and NT admins to perform
recovery tests. Hint, use Ignite on HP-UX to rebuild the box, make sure TSM
is installed on VG00, in its default directories.  Our NT Admins went
through restores following the Red Book and then experimented with their own
ideas.  They came up with some nice alternatives that proved to really kick
start the restores into high gear!  We also setup a small NT server and ATL
tape library that the UX and NT Admins use for testing, training and
expermenting (recently Win2K).  It made a world of difference getting their
emotional buy-in and taking ownership on TSM.

        4. Be very proactive on your business growth.  Our servers are
growing 40-50% a year for three years running and disk storage, doubling to
tripling every year (My EMC Sales Rep loves me!)  Our very slow NT restore
was due to fragmentation across too many tapes (320 to be exact) and took 4
days Yikes!

        5. Remember, it is still only tape and will only run at the maximum
speed you can get out of tapes. For true, super fast DR, look into some
real-time replication (SRDF for EMCs).  Also take a hard look at storing
business critical data on a high availability solution (we use both EMC and
NetApps).

        Mark Kovacs
        Purdue Pharma
        Stamford, CT.


> Lynn Sattler
> Dana Corp
> Toledo Oh
> lynn.sattler AT dana DOT com