On Wed, 5 Sep 2001 20:56:02 -0500, it was written:
>I am trying (or being forced to)defend my decision to use Tivoli over
>Arcserve.  What are my reasons again?  I have been pitching Tivoli and
>rolling it out with goos success to a Netware and NT environment that
>consists os about 4TB of data over about 85 nodes.  Everything was good
>untill I got to Exchange 2000 and everybody wanted message level restore
>like Arcserve and Commvault.  Also Microsoft has a bug that of course
>only affects Tivoli (memory leak).  If it sounds like I am wining then I
>probably am.
> 
>I guess my question is...   does anyone have any pros or cons for Tivoli
>over Arcserve?  My Exchange environment is a two server cluster on a
>Compaq SAN.  It has two storage groups right now one at 127g and the
>other at 192g.  There are plans  to break them out into smaller stores
>(around 50g each).

1. TSM uses Microsoft's Exchange APIs. As a result, the TDP for
Exchange client does not support account ("brick") level restores of
Exchange data. The advantage here is that when Microsoft tells your
Exchange admin to apply a patch to fix something-or-other on Exchange,
you can rest easy knowing that the patch won't screw around with TSM's
abilities. On the other hand, Arcserve doesn't use Exchange APIs;
rather, it uses Microsoft's Messaging APIs. This allow for
account-level restores of data. However--you remember that Exchange
patch? Odds are, it will cause Arcserve to choke and puke. (I won't
even get into how slow the [non-standard] use of the MS Messaging API
makes account restores, or how often Exchange data restored with
Arcserve ends up corrupted.)

2. Arcserve does not scale well; never has, and (it seems) never will.
I've not gotten a firm number of how many clients or how much data it
takes for scaling to rear its ugly head, but you'll see a marked
difference in performance as your client and data loads grow.

3. When Arcserve was owned by Cheyenne, they told a Houston client of
mine that, in order to recover the data locked in DLT cartridges that
an Arcserve bug prevented restoration thereof, my customer would have
to fork over $20K in order to fix the bug. When Arcserve was bought by
CA, they told another client of mine that data similarly locked in
*their* DLT cartridges would cost them $35K. Different
management--same story.

4. The very architecture of TSM (full once--incremental forever, along
with copygroups and internal database managment) makes it a superior
product. Complex? Yes. Fully featured? You betcha.

--
Mark Stapleton (stapleton AT berbee DOT com)
Mark Stapleton (stapleton AT berbee DOT com)