True it is not always the only or the best solution but I generally find
that better design (gig ethernet) of backbones and more tape drives work for
the majority of situations (hence the smiley at the end of my line).

I also liked the performance increase I got just from a software upgrade
when we went from 3.11 to 3.2 it was amazing. Depending on the server
between 60% and 100% better throughput. So not everything is in our control.

-----Original Message-----
From: Ravi Channavajhala [mailto:ravi.channavajhala AT csfb DOT com]
Sent: Tuesday, October 17, 2000 1:43 PM
To: McMurphy, Tim
Cc: veritas-bu AT mailman.eng.auburn DOT edu
Subject: RE: [Veritas-bu] Preemptive scheduling


On Tue, 17 Oct 2000, McMurphy, Tim wrote:

Tim>If your full backups are taking more than 24 hours its time to get more
and
Tim>faster hardware (ie network, tape drives, slaves etc). Management loves
to
Tim>hear that :)

This is frequently one of the arguments, I come across.  I'm in a
situation exactly as the original poster of this message.  My full backups
run for too long and my incrementals start while full bkups still in
progress.  I have a STK-9710 silo, and E-4500s as the backup servers and
backing up 300 machines (all servers) with data equalling 4-5 TB.  My
network setup ain't bad.  I would be hard pressed to believe that I need
more (OK my be Gigabit Ethernet, or trunking).

For the record, I activated MPX too.  Still, I haven't gotten a
satisfactory level of performance.  I have to try other options
like run Sun's SRM type stuff and figure out.  My argument is, just 
adding "better" hardware is never a full solution, at best partial. 

-ravi