That would match what I've heard most often.  While users of most other
backup products are seeing an average of 20:1 on most data, that ratio
relies on having full backups to compare against.  TSM's progressive
incremental backup of filesystems significantly reduces the commonality
that you'll achieve with filesystem backups.  However, you should
achieve the same commonality that other backup product users see on
database and application backups because you have full backups to
compare against.

VERY important: You will not see de-duplication of any kind if you're
only using your Data Domain box (or any other de-dupe target) as a disk
pool for caching.  If you're not leaving previous versions (or previous
full backups of databases) on the device, it won't have ANYTHING to
compare against, and you'll end up with 2:1, as some users have
mentioned.  (You get 2:1 because after any de-dupe stages, they also run
compression against data sent to their device.)  And since most de-dupe
products' pricing is based on getting a 20:1 or so de-dupe ratio, you're
kind of getting ripped off.  (This means that after you test what
de-dupe ratio you're going to get with your data, make sure you compare
the price of the de-dupe target against what you would pay for just a
regular disk array.  If you're getting a very low de-dupe ratio, they
may be very close in price.)

Finally, if you're OK with just getting 2:1, you might consider
Storewiz.  They are a gateway in front of your storage that does in-line
Limpel-Ziv compression on your data allegedly without any performance
degradation.  (The idea is that while compressing obviously takes some
cycles, they make up those cycles by having your disk array store 50%
fewer blocks.)

---
W. Curtis Preston
Backup Blog @ www.backupcentral.com
VP Data Protection, GlassHouse Technologies 

-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Cory Heikel
Sent: Wednesday, October 10, 2007 6:38 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: [ADSM-L] TSM and Data Domain

We are averaging about 11/1 with data domain on a mix of data. One thing
we found was the need to update a setting for tsm tape marker parameter,
which almost doubled our performance.
 
 
Cory Heikel
Tivoli Systems Administrator
Hershey Medical Center
(717) 531-7972

>>> Dirk Kastens <Dirk.Kastens AT UNI-OSNABRUECK DOT DE> 10/10/2007 2:45 AM >>>
Kenneth,

Bradberry, Kenneth schrieb:
> I am currently looking at Data Domain as a possible VTL and
> de-duplication vender.  Can anyone share there experiences with Data
> Domain with TSM?

We just tested a DataDomain machine with TSM. The results were not as
good as we expected. It shows a very good compression with database
backups (Exchange backups should also work well) but not with normal
user data, like homedirectories or web data.

--
Regards,

Dirk Kastens
Universitaet Osnabrueck, Rechenzentrum (Computer Center)
Albrechtstr. 28, 49069 Osnabrueck, Germany
Tel.: +49-541-969-2347, FAX: -2470