ADSM-L

Re: [ADSM-L] TSM dream setup

2008-02-15 21:24:53
Subject: Re: [ADSM-L] TSM dream setup
From: Roger Deschner <rogerd AT UIC DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Fri, 15 Feb 2008 20:24:22 -0600
.
We did a financial analysis of dedpue with TSM vs. just simply buying
more disks, and the more disks solution came in cheaper. Plus a huge
advantage in simplicity. Those dedupe appliances are expensive! Their
sales pitch is that they can compress 20:1, so they charge 12 times as
much money. But actual dedpue ratios with TSM incremental data are more
like 5:1 to 10:1, so dedupe costs MORE. Dedupe is coming as a feature in
TSM v6 anyway, so don't get too excited about 3rd party solutions yet.

Back to dream systems, we are presently bringing in a major upgrade.
(After a slight delay due to our recent disaster recovery situation. BTW
Thanks again to all on this list who helped! Your insight and ideas were
very valuable.)

We'll have two systems. Our original, running on a venerable but
reliable RS/6000 H80, will trade in its venerable but UNreliable Quantum
P7000 SCSI library with SDLT320 drives for a Sun STK SL500 library with
LTO4 drives. This will continue to back up the big server systems, which
have always been ideally suited to tape collocation. Therefore backup to
tape secondary stgpools is actually still better than backup to disk.
(Also better than non-TSM backup to local tape.) For primary stgpools
and database it has a mix of nearly every size, speed, and color SSA
disk ever made. It wasn't planned; it happened incrementally.

The new system, which will back up the thousands of desktop systems that
are too small to collocate, or to group collocate effectively without a
lot of work, will be a disk-only system. It will be a used pSeries 6M2
running AIX. It's "library" will be a new Sun X4500 "Thumper" with
48tb(raw) of SATA disk running ZFS (RAID-Z). This will totally eliminate
the issue of non-collocation we had in our recent disaster. Copy stgpool
will be in the SL500 library on the other system. It will have a drawer
of 36gb 15,000rpm IBM SSA disks running in JBOD mode for the TSM
database, and a drawer of 72gb 10,000rpm SSA disks for primary storage
pools. (Hey - our used eqp dealer says SSA is the best buy in used
high-performance disks, especially if you already know how to run them,
which I do.) The two systems are physically separated by 1.5 miles in
case of calamity, but connected by our own private dark fibre.

So that's my dream medium-sized medium-budget TSM system, and I'm
finally getting it. It wasn't all that expensive. (Couldn't get any
high-budget solution, because the State of Illinois is broke.) I like
that it's TWO systems. And it's going to have enough room for the sudden
growth of interest in TSM backup after that recent disaster. Those who
backed up to TSM are now saying to those who weren't, "I have my data
back, do you?"

Roger Deschner      University of Illinois at Chicago     rogerd AT uic DOT edu
               Academic Computing & Communications Center
============ "In theory, theory and practice are the same, =============
========= but in practice, theory and practice are different." =========

On Thu, 14 Feb 2008, Stapleton, Mark wrote:

>> I do have a question on something someone mentioned and that is de-
>> deduplication.  Are many of you using it and do you consider it an
>extra
>> level on potential problems?  Just curious.
>
>It's highly overrated with TSM, since TSM doesn't do absolute (full)
>backups unless such are forced. If you want maximum bang for your buck
>with dedupe, you need to run an app like IBM CommonStore for Email that
>will enable SIS (single-instance store) on your email server's data.
>
>When a CEO sends a 4MB attachment to 100 different email addresses, it
>makes mail administrators cry.
>
>--
>Mark Stapleton
>Berbee (a CDW company)
>System engineer
>7145 Boone Avenue North, Suite 140
>Brooklyn Park MN 55428-1511
>763-592-5963
>www.berbee.com
>

<Prev in Thread] Current Thread [Next in Thread>